Compare commits

...
Author SHA1 Message Date
jschoubben 048f1b8284 ssh-client module: the mesh owns ~/.ssh, config from the hub (to-be 29)
Requires openssh; creates ~/.ssh (0700, owned by the operator account via
${machine:account}); writes every other node's Host block (HostName + User
<account>) into a marked region of ~/.ssh/config (home-scoped, into:block), so
`ssh <node>` reaches each peer as the right account and the operator's own
config is kept. Universal-tier: assigned wherever a person logs in; a node with
no account gets no config.
2026-09-27 17:50:57 +02:00
jschoubben 278610c0c3 fail2ban never bans a tunnel peer: ignoreip names the mesh range
The jail.local [DEFAULT] gains ignoreip = 127.0.0.1/8 ::1 ${machine:mesh-range}
— localhost plus the mesh's own private range, named through the placeholder
rather than hardcoded (data is the mesh's, ADR 0112). Without it fail2ban could
ban the mesh's own nodes on 10.10.0.0/24; on novox that rule survived only in
memory from a now-deleted HAL file and would be lost on the next restart.
2026-09-27 16:55:34 +02:00
jschoubben 6bedcd3f21 Rename seat claims to the mesh-*/node-* convention; retire verdaccio (ADR 0121)
Claims renamed to match the controller's seat set: node-dns-resolver (dnsmasq),
node-intrusion-prevention (fail2ban), node-packet-filter (nftables),
node-resolver-config (resolv-conf, resolved-split-dns), node-uplink
(networkmanager, systemd-networkd, dhcpcd), mesh-build-machine (builder, +mesh
scope), mesh-catalog (mesh-catalog). showcase now declares its own seat and
claims it. verdaccio removed — the mesh keeps distribution as its registry and
gitea already serves npm, so a second npm registry is redundant.
2026-09-27 14:30:56 +02:00
jschoubben 968473219a dnsmasq owns its resolver format: node-zones is a template, not a controller formatter (ADR 0120)
The node-zones fact was a path; the local=/address= syntax lived in the
control plane. It is dnsmasq's configuration language, so it moves into
dnsmasq's manifest as a template over the roster. The mesh renders it; it
reads none of it. Output is unchanged.

Lands with mesh-controller's ADR 0120 change — the two are one schema step.
2026-09-27 01:33:28 +02:00
jschoubben ad219beee2 Merge pull request 'The uplink's managers are modules: networkmanager, systemd-networkd, dhcpcd (hq ADR 0117)' (#110) from feat/the-uplink-modules into main 2026-09-26 23:00:30 +00:00
jochen fd09b1a50e review: the uplink managers are the machine's — no state on their services, none for dhcpcd; comments corrected 2026-09-27 00:07:27 +02:00
jochen 7aea08d6c3 dhcpcd's mesh block goes at the start: lines after an interface line are that interface's 2026-09-26 23:48:20 +02:00
jochen 23d735a0bf The uplink's managers are modules (hq ADR 0117)
networkmanager, systemd-networkd and dhcpcd each claim the-uplink and
declare only what keeps the machine's own network manager from
contradicting the mesh: the resolver file left to resolv-conf, mesh0
left alone. Never a link, profile or credential — the link is the
mesh's only channel to the machine, so NetworkManager and networkd are
reloaded on a change, never restarted, and dhcpcd (no reload; a restart
drops the address) takes its block at its next start.
2026-09-26 23:47:31 +02:00
jschoubben 226eab4c6f Merge pull request 'dnsmasq: the operator's own names have a home the mesh never rewrites' (#109) from feat/dnsmasq-has-a-home-for-operator-names into main 2026-09-26 20:43:53 +00:00
jschoubben bb8f2e76a9 dnsmasq: the operator's own names have a home the mesh never rewrites
A workstation's job includes names that are neither a mesh machine nor
a routed name (novox/hq 122): shanks carries 13 Mediahuis entries in
/etc/hosts, and mesh-wireguard replaces /etc/hosts whole when taken —
so without this they vanish, and the take gates the node. Two homes,
neither the mesh's to own: conf-dir=/etc/dnsmasq.d/,*.conf (drop-in
directives, HAL's dnsmasq-app used exactly this) and
addn-hosts=/etc/hosts.local (plain host lines). The mesh creates and
rewrites neither; a machine with none loses nothing. The migration
moves such names here BEFORE the /etc/hosts take, closing the window.
2026-09-26 22:43:29 +02:00
jschoubben 372450851f Merge pull request 'mssql: its data is placed — the last /services placement retires' (#108) from feat/mssql-data-is-placed into main 2026-09-26 18:36:16 +00:00
jschoubben f4e4e12c99 mssql: its data is placed — the last /services placement retires
The stated path was the adopted-data exception; with the take done and
the placement vocabulary live, the exception has no reason left. The
landing window renames the directory and recreates the container, since
a changed volume path does not do that by itself (hq 126).
2026-09-26 20:36:03 +02:00
jschoubben fd9be011c0 Merge pull request 'sshd: the operator's door is a module' (#107) from feat/sshd-module into main 2026-09-26 18:15:08 +00:00
jschoubben a85b0ee346 sshd: the operator's door is a module
The spec is the working system: HAL's 99-hal.conf, restated as
10-mesh.conf so lexical include order makes the mesh's answer the one
that wins while the predecessor's file is still on disk. Subsystem
stays the stock config's — first-set wins and it sits before the
Include. Port 22 from anywhere, said in listens with its reason: the
machines that need the door are exactly the ones not on the mesh yet,
and locking the operator out is the one failure a firewall must never
arrange.
2026-09-26 20:14:54 +02:00
jschoubben 34243c9e34 Merge pull request 'dnsmasq: the runtime's DNS is written into daemon.json, never over it' (#106) from fix/dnsmasq-writes-into-daemon-json into main 2026-09-26 18:12:20 +00:00
jschoubben 870a541072 dnsmasq: the runtime's DNS is written into daemon.json, never over it
The file is shared — the operator's insecure-registries for the mesh's
own store live there — and replacing it whole would break every pull
from that store the moment the module is taken (the 098 class, caught
in the pre-take diff this time). ADR 0102's verb is merge.
2026-09-26 20:12:06 +02:00
jschoubben a59750fa28 Merge pull request 'mailu: the smtp provision serves the name its certificate answers to' (#105) from fix/smtp-serves-its-tls-name into main 2026-09-26 17:27:06 +00:00
jschoubben 81592a3b2c mailu: the smtp provision serves the name its certificate answers to
A consumer connecting by the binding's address meets a certificate for
mail.novox.be and refuses it — found live by the forwarder's cutover
proof, one send before production would have. The TLS name is mailu's
own fact (HOSTNAMES), so the binding carries it; consumers say
${bound:smtp:name} and verification holds.
2026-09-26 19:26:53 +02:00
jschoubben 705ceec1e7 Merge pull request 'Six modules name no /var/lib: the root is a place, the maps reference it' (#104) from feat/six-modules-name-no-var-lib into main 2026-09-26 16:21:00 +00:00
jschoubben afdd149ab7 Six modules name no /var/lib: the root is a place, the maps reference it
The state directories say place "." — the assignment's own root — and
every bind, secret, own-secret, receives and grants path references it
as ${dir:state}/…; grants directories that are their own resources are
placed by id. gitea's two coincidence strings from the first pass
(${dir:data}base.json — resolving correctly by pure concatenation) are
spelled honestly now. What still says /var/lib is inside containers —
the software's contract — or under /var/lib/mesh, the mesh's own
plumbing, which the requirements unification absorbs next. Every
resolved path is byte-identical to what runs; landing this is a no-op
on the node, and the converter checks its own boundaries this time.
2026-09-26 18:20:47 +02:00
jschoubben dde8b15483 Merge pull request 'mailu: seventeen data directories are placed, not stated' (#103) from feat/mailu-dirs-are-placed into main 2026-09-26 16:07:24 +00:00
jschoubben 8a046be198 mailu: seventeen data directories are placed, not stated
Each resolves to <root>/mailu/<id> — the maildir at
/var/lib/mailu/data-mail, certs at data-certs, and so on. Landing this
is a window, not an edit: seventeen renames on the node (the nested
data/ tree flattens to the ids), then the full stack recreated, because
a changed volume path does not recreate a container by itself (hq 126).
Ids are untouched on purpose — a renamed id orphans its held record,
and the mail spool is the wrong place to learn what a removal step does
with one.
2026-09-26 18:07:11 +02:00
jschoubben 668278bde4 Merge pull request 'nextcloud: it lives under its own name, and html is placed' (#102) from feat/nextcloud-lives-under-its-own-name into main 2026-09-26 16:05:41 +00:00
jschoubben 6761bb02a1 nextcloud: it lives under its own name, and html is placed
The module is nextcloud; its tree was /var/lib/nextcloud-module — a
historic spelling nothing depends on. The root moves to
/var/lib/nextcloud (a rename on the node, done in this change's
window), and html drops its path: the mesh resolves it to
<root>/nextcloud/html. Landing this requires the window: rename the
tree, push, recreate the container — a changed volume path does not
recreate one by itself (hq 126).
2026-09-26 18:05:28 +02:00
jschoubben f98c9859d2 Merge pull request 'gitea: its data and grants are placed, not stated' (#101) from feat/gitea-dirs-are-placed into main 2026-09-26 16:04:30 +00:00
jschoubben cac5eab7da gitea: its data and grants are placed, not stated
The mesh resolves both to <root>/gitea/<id> — where the 5.8G forge and
its grant files already sit, so the roll-out its upgrade policy makes
of this build changes no byte of the spec. The module root and the
mesh's plumbing stay stated.
2026-09-26 18:04:17 +02:00
jschoubben 770c9f6a78 Merge pull request 'mongodb: its data directory is placed, not stated' (#100) from feat/mongodb-dir-is-placed into main 2026-09-26 16:03:36 +00:00
jschoubben 5427118614 mongodb: its data directory is placed, not stated
The mesh resolves it to <root>/mongodb/data — where the granted
databases already sit. The provider's own state, grants and the mesh's
plumbing stay stated.
2026-09-26 18:03:25 +02:00
jschoubben 53765335cf Merge pull request 'portainer: its data directory is placed, not stated' (#99) from feat/portainer-dir-is-placed into main 2026-09-26 16:02:42 +00:00
jschoubben 5fd2ed9686 portainer: its data directory is placed, not stated
The mesh resolves it to <root>/portainer/data — where the 16M of
endpoints and users already sit. A textual no-op on this node.
2026-09-26 18:02:25 +02:00
jschoubben f7887d706d Merge pull request 'only-office: its directories are placed, not stated' (#98) from feat/only-office-dirs-are-placed into main 2026-09-26 16:01:43 +00:00
jschoubben d6dd21a091 only-office: its directories are placed, not stated
Seven data directories drop their paths; the mesh resolves each to
<root>/only-office/<id>, which is exactly where the data already sits —
a textual no-op on this node, and the first module speaking ADR 0112's
vocabulary. The module root and the mesh's own state stay stated.
2026-09-26 18:01:31 +02:00
jschoubben a844701577 Merge pull request 'Module data lives in /var/lib, now that nothing is mid-cutover' (#97) from feat/module-data-lives-in-var-lib into main 2026-09-26 14:47:37 +00:00
jschoubben 50a99f022c Module data lives in /var/lib, now that nothing is mid-cutover
The /services paths were the adopted-node pattern doing its job: take
replaced containers over the predecessor's data without moving a byte
(gitea set it — 'its data never moved'). With every cutover done the
exception has no reason left, and the operator called it: a nox
module's world is /var/lib/<module>, data included. Six modules
repathed; mssql keeps its /services path deliberately — it is still
held, HAL-run, and moves at its own take. Both trees are one
filesystem, so each move is a rename.
2026-09-26 16:47:24 +02:00
jschoubben 382a44621e Merge pull request 'A bucket is the one the mesh derives, and the photos module named another' (#96) from fix/a-bucket-is-the-one-the-mesh-derives into main 2026-09-26 14:46:30 +00:00
jochen ddb67fc095 A bucket is the one the mesh derives, and the photos module named another
The provisioner derives a consumer's bucket from the login the mesh minted — 'derived from the
login, so teardown recomputes it with nothing to persist' — and never reads the bucket a manifest
contributed. Three modules contributed one anyway, and the value was decorative in two and wrong in
the third: photos told its container MINIO_BUCKET=photos, the predecessor's bucket, while its minted
key is scoped to mesh-novox-photos. Deployed as it stood, it would have authenticated and then been
denied on every object.

photos now names the bucket the mesh actually provisions, and the contributed bucket is gone from
all three: a value nothing reads, that reads as though it decides.

Verified against the live store before changing anything: the derived names are the populated ones —
mesh-novox-ncloud (77,886 objects, 174.9 GiB), mesh-novox-photos and mesh-novox-invoice. Nothing has
to move.
2026-09-26 16:46:10 +02:00
jschoubben 78595e4db3 Merge pull request 'lavinmq: the broker TLS directory is the operator's, read by whoever needs it' (#95) from fix/the-broker-tls-directory-is-the-operators into main 2026-09-26 14:12:45 +00:00
jschoubben 37634de1e3 lavinmq: the broker TLS directory is the operator's, read by whoever needs it
The controller now accesses /var/lib/mesh-broker-tls (mesh-controller
#54) and the push refused whole: lavinmq declared the directory as an
owned resource, and shared data is the operator's, owned by no module
(ADR 0051). lavinmq only ever reads the certs — genesis laid them down
— so it declares a read access like the controller does, and the
directory belongs to nobody.
2026-09-26 16:12:29 +02:00
jschoubben 36c5f87130 Merge pull request 'route-adapter: write a body limit as the predecessor's buffering middleware' (#94) from feat/a-route-may-limit-the-body-it-carries into main 2026-09-26 14:02:03 +00:00
jochen b2e39eb2cd route-adapter: write a body limit as the predecessor's buffering middleware
The adapter skips what its one file shape cannot say. A body limit is the exception: the predecessor
has a buffering middleware and served its own registry name with exactly it, so this is written
rather than skipped, named after the router so the two halves cannot drift.

A limit that is not a whole positive number of bytes takes the route with it. Written without the
limit, the predecessor would carry what the module said not to carry and this module would report
success. Silence stays silence — no middleware, the predecessor's default.
2026-09-26 16:01:23 +02:00
jschoubben 3d73c9f54e Merge pull request 'nextcloud: real mesh module, MariaDB→PostgreSQL, S3 via _FILE secrets' (#59) from feat/nextcloud-module-postgres-migration into main 2026-09-26 13:06:32 +00:00
jschoubben fb95eb6e46 Merge main 2026-09-26 15:06:11 +02:00
jschoubben 4d715f8b73 Merge pull request 'minio: the real 4-node/8-drive erasure-coded cluster, both public routes, verified live on novox' (#58) from feat/minio-real-cluster-not-single-node into main 2026-09-26 13:05:57 +00:00
jochen afe8aae826 Merge main
# Conflicts:
#	modules/minio/module.json
2026-09-26 15:05:40 +02:00
jschoubben fa91be4941 Merge pull request 'postgres: declare the data directory's real owner; keycloak: use the port template' (#55) from fix/postgres-owner-and-keycloak-port-template into main 2026-09-26 13:05:10 +00:00
jschoubben 87f73dce6a Merge main 2026-09-26 15:04:47 +02:00
jschoubben fc5ccdfe2a Merge pull request 'mailu: one WEBMAIL_ADDRESS, the mesh's container name' (#93) from fix/one-webmail-address into main 2026-09-26 13:04:44 +00:00
jschoubben 4489e56935 mailu: one WEBMAIL_ADDRESS, the mesh's container name
The env block carried the key twice — mailu-webmail from #79's address
sweep, and a stray =webmail further down that survived it. Last write
wins in an env file, so the front resolved a name that answers nowhere
on the mesh's network and 502'd every logged-in webmail request. Latent
since the cutover: the SSO redirect the checks watched never touches
the upstream; the operator's first real login did.
2026-09-26 15:04:32 +02:00
jschoubben 08e947e4c8 Merge pull request 'The npm registry is a seat gitea holds, and gitea holds the git seat a build's source can live on' (#69) from feat/seats-are-a-closed-set into main 2026-09-26 12:31:10 +00:00
jschoubben 7fb9dd0254 Merge main 2026-09-26 14:29:28 +02:00
jschoubben 420d05e8dd Merge pull request 'mailu certifies itself, take two — the fall-through is now a behaviour' (#92) from fix/mailu-certifies-itself-take-two into main 2026-09-26 12:27:56 +00:00
jschoubben 6769e66c82 mailu certifies itself, take two — the fall-through is now a behaviour
Take one (#90) died on two real edge bugs, both fixed and pinned by
tests in mesh-controller (#66: autocert 404s unknown tokens itself;
#67: the internal authority 403s every public name before the token
lookup). The challenge path verified end to end reaching mailu's own
nginx before this flip.
2026-09-26 14:27:45 +02:00
jschoubben 15b35b53db Merge pull request 'mailu: back to the copied cert — the edge's fall-through is a belief, not a behaviour' (#91) from fix/mailu-back-to-cert-while-the-fallthrough-is-fixed into main 2026-09-26 12:19:17 +00:00
jschoubben 6177565741 mailu: back to the copied cert — the edge's fall-through is a belief, not a behaviour
The letsencrypt flavor served certbot's April-expired state to live IMAPS
users within minutes: autocert's HTTPHandler answers 404 itself for
tokens it does not hold and never consults the fallback for challenge
paths, so mailu's own client cannot answer through the path-scoped
route. cert flavor (valid to Nov 27) until route-proxy's handler
actually falls through.
2026-09-26 14:19:05 +02:00
jschoubben 6b2ea0972a Merge pull request 'mailu certifies itself: the edge passes unknown ACME tokens through now' (#90) from fix/mailu-certifies-itself into main 2026-09-26 12:15:24 +00:00
jschoubben a978b53d1c mailu certifies itself: the edge passes unknown ACME tokens through now
PR #82 set TLS_FLAVOR=cert as the honest interim while the predecessor's
proxy owned /.well-known/acme-challenge outright. route-proxy took port
80 today and its handler passes unknown tokens through to routed paths
by design — the one line #82 promised, made now. The copied cert (valid
to Nov 27) stays on disk untouched; mailu's own certbot takes over from
here.
2026-09-26 14:15:12 +02:00
jschoubben 7501c1db9e Merge pull request 'portainer: serve its public name, hold its real data, run the image the machine runs' (#89) from fix/portainer-serves-its-name into main 2026-09-26 01:46:24 +00:00
jschoubben 00ada1e9f7 portainer: serve its public name, hold its real data, run the image the machine runs
The manifest predated the working deployment on three axes: it declared a
data directory the running portainer never used (taking it would have
started empty), pinned an image digest the machine has moved past (issue
099), and contributed no route while portainer.novox.be rides a traefik
container label today. Now: the predecessor's portainer_data path, the
running image's digest, 9090:9000 kept as the predecessor's machine port
with the route contribution naming it, and 9443 kept for the runtime
sidecar's own TLS conversation.
2026-09-26 03:46:11 +02:00
jschoubben 67b443d5ad Merge pull request 'only-office: pin the machine side of its port' (#88) from fix/only-office-pins-its-machine-port into main 2026-09-26 01:44:59 +00:00
jschoubben a2da2e4910 only-office: pin the machine side of its port
A bare '80' tried to bind the node's port 80 — the edge's — instead of
auto-allocating. 9070 is the predecessor's number and the one the route
contribution already names.
2026-09-26 03:44:47 +02:00
jschoubben bcb9ca8f93 Merge pull request 'invoicing: MONGO_DB says the granted database's name' (#87) from fix/invoicing-names-its-database into main 2026-09-26 01:34:09 +00:00
jschoubben 2409afda60 invoicing: MONGO_DB says the granted database's name
The app reads MONGO_DB (default 'invoicing') for every operation and
uses the URL only to connect — listCollections ran against a database
the granted user cannot see. Same fault and same fix as photos' MONGO_DB,
found by the API's own logs at take.
2026-09-26 03:34:00 +02:00
jschoubben 511200ed9c Merge pull request 'invoicing: the photos lessons, applied before its window' (#86) from fix/invoicing-learns-the-photos-lessons into main 2026-09-26 01:15:38 +00:00
jschoubben b704bf5ad8 invoicing: the photos lessons, applied before its window
The mongo credential authenticates against its own database and the
database is the granted one (mesh_novox_invoice), not the contributed
name the provisioner ignores. Same for the store: the key is sealed to
the derived bucket (mesh-novox-invoice) — the data mirrors in during the
window, the ncloud/photos pattern. And the api gets the route
contribution it always needed: invoicing-api.novox.be is today a traefik
container label, invisible to every file survey, and it must be a grant
before the edge can ever flip.
2026-09-26 03:15:26 +02:00
jschoubben 142d65c52a Merge pull request 'mongodb: the server container is mongodb-server, not the predecessor's name' (#85) from fix/mongodb-coexists-with-the-predecessor into main 2026-09-26 00:37:51 +00:00
jschoubben ccb6e7500e mongodb: the server container is mongodb-server, not the predecessor's name
The adopted node still runs the predecessor's mongo container, and it must
keep running: invoicing points at novox.be:27017 and is not migrating in
this window. A module container named 'mongo' would be held at assign and
would replace the predecessor at take, cutting invoicing off its database.
The mesh's server coexists instead — fresh data directory, its own name,
auto-allocated machine port — and the predecessor retires with its last
consumer.
2026-09-26 01:37:41 +02:00
jschoubben bbda88c13b Merge pull request 'Every credential provider says whether it still holds a consumer (hq issue 120)' (#84) from fix/120-redis-says-what-it-holds into main 2026-09-25 23:31:28 +00:00
jochen 620b47d309 mssql remaps a user only when orphaned; mailu's operator tool no longer re-enables
ALTER USER ... WITH LOGIN runs only when the user's SID is not the
login's, so an already-mapped user is left alone. The provisioner
enables a mailbox through its own method; the password tool an
operator uses keeps changing the password only.
2026-09-26 01:30:15 +02:00
jochen 6fd93afc6c Review fixes: holds and create agree, and no password leaves a check
create re-enables what holds refuses (mssql login, mosquitto client,
mailu mailbox, gitea user) and clears an expired postgres password, so
no disabled account loops. mssql and mongodb checks take the password
from the environment, never argv; mosquitto_ctrl failures no longer
repeat -P. mosquitto reads 'could not ask' as an error, not absence.
mailu checks existence and enabled only: its imap passdb cannot verify
a password. mssql checks the user's SID; gitea pages teams at 50.
2026-09-26 01:24:32 +02:00
jochen 0cb0f814b4 Every credential provider says whether it still holds a consumer
holds() for postgres, mssql, mongodb, minio, lavinmq, mosquitto, mailu
and gitea, so the harness makes again a login the backend lost (hq issue
120). Each checks the mesh's password as the consumer presents it, or
compares it read-only, and returns false only when the backend says the
credential is absent or wrong; an unreachable backend throws.
2026-09-26 01:09:52 +02:00
jochen 3d271f72ea redis: say whether it still holds a consumer's ACL user
The server keeps ACL users in memory only, so a restart forgets every
consumer while the provisioner keeps running (hq issue 120). holds()
checks ACL GETUSER for the user, enabled, with the mesh's password, so
the harness makes a forgotten user again. Needs mesh-sdk 0.1.1.
2026-09-26 00:53:13 +02:00
jschoubben 76ca479f37 Merge pull request 'mailu: the queue is root's and traversable, which is postfix's own convention' (#83) from fix/the-queue-is-traversable into main 2026-09-25 22:17:54 +00:00
jschoubben a055334c9b mailu: the queue is root's and traversable, which is postfix's own convention
The declared 0700 was applied at take and broke mail quietly: postfix's
master runs as root but pickup and smtpd drop to uid postfix, and a
spool root they cannot traverse is a maildrop they cannot scan and a
rewrite socket they cannot open — auth succeeded and MAIL FROM hung.
0755 root is exactly what postfix's own set-permissions makes of
/var/spool/postfix. Fixed live by chmod first; declared here so the
next push stops undoing it.
2026-09-26 00:17:41 +02:00
jschoubben d5b169b6b6 Merge pull request 'mailu: cert flavor while the predecessor's proxy owns the challenge path' (#82) from fix/mailu-tls-cert-behind-the-predecessors-proxy into main 2026-09-25 22:11:26 +00:00
jschoubben 8ecc5a7249 mailu: cert flavor while the predecessor's proxy owns the challenge path
letsencrypt was the aspiration and cannot work yet, proven live: the
predecessor's own ACME machinery owns /.well-known/acme-challenge on
port 80 outright (unknown tokens get its 404) and its entrypoint
redirect owns every other path — the hand-authored passthrough never
matched anything, which is why mailu's certbot state had quietly
expired in April while the copied files carried the name. cert flavor
serves those files (valid to Nov 27). Mailu certifying itself becomes
possible the day route-proxy takes port 80, whose handler falls through
unknown tokens by design — that flip is one line here, made then.
2026-09-26 00:11:11 +02:00
jschoubben 29f00f33a5 Merge pull request 'automx: the seed writes the schema 2021.6 reads' (#81) from fix/automx-seed-knows-prio into main 2026-09-25 22:06:36 +00:00
jschoubben 480627fdd9 automx: the seed writes the schema 2021.6 reads
The hand-written seed predates automx2's prio column, so every
config-v1.1.xml request 500'd against a table the seed had just made —
and the predecessor's own database had the same gap: client
autoconfiguration has been silently broken on the old stack for a long
time, behind a root page that answered 200. The live database gained
the column by ALTER; a fresh mesh now seeds it right.
2026-09-26 00:06:24 +02:00
jschoubben c8c939a595 Merge pull request 'automx: the schema its seed writes belongs to one automx2, so that one is named' (#80) from fix/automx-pins-the-schema-its-seed-writes into main 2026-09-25 21:59:54 +00:00
jschoubben 30b7ce429c automx: the schema its seed writes belongs to one automx2, so that one is named
An unpinned pip install took the latest automx2, whose schema grew a
column (server.prio) the module's own seeding SQL predates — a 500 on
every autoconfig request against a table the seed had just written.
2021.6 is what the proven image runs; the seed and the software agree
again. Regenerating the seed for a newer automx2 is its own change,
made deliberately, not by whatever pip resolved this week.
2026-09-25 23:59:42 +02:00
jschoubben b8a50ee7a0 Merge pull request 'mailu: the containers are named by the vocabulary 2024.06 reads' (#79) from fix/mailu-speaks-2024-06-addresses into main 2026-09-25 21:56:40 +00:00
jschoubben 4e37b3e84d mailu: the containers are named by the vocabulary 2024.06 reads
The front resolves its upstreams from *_ADDRESS, defaulting to the bare
compose service names — admin, antispam — and reads the 1.9-era HOST_*
not at all. Phase 1 never noticed because the predecessor's service
names WERE the defaults; the mesh's containers are mailu-*, and the
front answered 502 asking docker for a name nothing carries. The dead
vocabulary goes; every upstream is named as the container actually is.
2026-09-25 23:56:28 +02:00
jschoubben 7cb7659fbe Merge pull request 'mailu: every container asks the module's own resolver, pinned where they can find it' (#78) from feat/mailu-points-at-its-own-resolver into main 2026-09-25 21:49:30 +00:00
jschoubben 047f228fe3 mailu: every container asks the module's own resolver, pinned where they can find it
2024.06's admin refuses to serve behind a resolver that does not
validate DNSSEC — found live as an unhealthy admin, 454s on submission
and a 500 webmail, with the runtime's forwarder validating nothing. The
unbound this module always shipped becomes reachable: pinned at the
predecessor's own address on the module network (the subnet the mesh
adopted), and named as dns by the nine containers that resolve anything.
Stands on mesh-host #26, which gave the vocabulary these two fields.
2026-09-25 23:49:17 +02:00
jschoubben 40aa93dc2b Merge pull request 'mailu: the admin API answers on 8080 since 2024.06' (#77) from fix/mailu-admin-answers-on-8080 into main 2026-09-25 21:45:11 +00:00
jschoubben 5462317183 mailu: the admin API answers on 8080 since 2024.06
1.9's admin served on 80; 2024.06's gunicorn listens on 8080, and the
runtime's fetch failed against the old port the moment the broker
credential let it try.
2026-09-25 23:45:00 +02:00
jschoubben cb9d43b2a1 Merge pull request 'automx: the launcher's wrapper is written by the build that ships it' (#76) from fix/automx-carries-its-own-wrapper into main 2026-09-25 21:44:46 +00:00
jschoubben aa8c4253f1 automx: the launcher's wrapper is written by the build that ships it
The predecessor's image carried .venv/scripts/flask.sh, created by a
build step that never made it into the files this module holds — the
image worked and its recipe could not reproduce it, caught the moment
the mesh built it from source (exit 127 crash loop at cutover). The
wrapper is now written explicitly, verbatim from the proven image, so
the recipe is the whole truth about the image.
2026-09-25 23:44:34 +02:00
jschoubben 4863eef586 Merge pull request 'mailu: pin exactly what phase 1 verified' (#75) from fix/mailu-pins-what-phase-one-verified into main 2026-09-25 21:34:26 +00:00
jschoubben 75f993e92b mailu: pin exactly what phase 1 verified
Phase 1 (MAILU-CUTOVER.md) upgraded the live stack 1.9→2024.06 and its
lessons land here as pins: every image is the digest running and
verified tonight — webmail under the name 2024.06 actually uses (the
draft's roundcube pin misled a whole hop), antivirus on the upstream
clamav image with the signature DB in its own directory (the mailu-built
image ended at 2.0), and the front trusting the proxy address the live
config actually names. The welcome-mail texts ride along for parity.

TLS_FLAVOR stays letsencrypt deliberately where the live .env says cert:
the cert files are copies whose HAL-era renewal hook died with HAL
(expiry Nov 27); mailu managing its own issuance through the existing
ACME passthrough is the fix, and the files remain on disk as the
fallback flavor if first issuance misbehaves during the window.
2026-09-25 23:34:13 +02:00
jschoubben 621246996d Merge pull request 'mailu: 2024.06 closes 110/143/587 by default; parity says open them' (#74) from fix/mailu-ports-parity into main 2026-09-25 20:56:39 +00:00
jschoubben e3d8bd0726 mailu: 2024.06 closes 110/143/587 by default; parity says open them
PORTS defaults to 25,80,443,465,993,995,4190 in 2024.06 — submission on
587 among the closed, which is what every client of this server uses.
The same parity decision the listens already state, now stated where the
software reads it.
2026-09-25 22:56:27 +02:00
jschoubben a708666bad Merge pull request 'mailu: the manifest matches the machine, provides smtp, and carries automx' (#73) from feat/mailu-becomes-real into main 2026-09-25 20:40:08 +00:00
jschoubben ed5d1386ce mailu: the manifest matches the machine, provides smtp, and carries automx
Five gaps between the draft and what actually runs, each verified live
before being written down:

- front published bare 80 — the machine port Traefik holds; now the
  predecessor's own mappings (7080:80, 7443:443) plus the 110/143/995
  parity ports the draft dropped. Pruning legacy protocols is its own
  deliberate change, not a cutover side effect.
- TLS_FLAVOR said cert, which nothing supplies; live is letsencrypt —
  mailu runs its own certbot, state already on disk, HTTP-01 answered
  through a path-scoped route contribution (priority above the web one).
- the web route said http:7080, the redirect-loop shape; it now says
  what the hand-authored file always knew: https 7443, insecure.
- automx was absent entirely: the autoconfig responder is now a second
  artifact (its Containerfile moved in from the predecessor's images
  dir, base declared per ADR 0097), a container on a real data dir —
  the anonymous-volume loss of 2026-08-10 stays fixed — and the three
  public names are route contributions.
- and the reason this moved ahead of de-spiegel: mailu now provides
  smtp. A consumer contributes the account it sends as; the provisioner
  creates <account>@<domain> via the admin API and applies the minted
  password every reconcile (ADR 0048). The domain is served on the
  binding so a consumer composes its own login from mesh facts.

route-adapter learns to say no: a contribution over https, scoped to a
path, or carrying a policy is skipped aloud rather than written into a
file shape that cannot say it — plain http into a TLS listener was the
concrete wrong file this prevents. The hand-authored files keep covering
those routes until the mesh's own proxy takes over, exactly as today.
2026-09-25 22:39:29 +02:00
jschoubben 3875987656 Merge pull request 'gitea: the package team may read code, and its units are reconciled' (#72) from fix/gitea-package-team-reads-code into main 2026-09-25 19:59:25 +00:00
jschoubben 311f7f1fdb gitea: the package team may read code, and its units are reconciled
The builder's first credentialed clone of a private repository answered
'not found': the packages team named only repo.packages in its
units_map, which is exhaustive — so members had no code unit at all, and
gitea hides what a user cannot read. One credential answering npm and
git alike was the whole design of the builder's grant; the team now says
so.

And found teams are patched, not just returned: a team is configuration
the reconcile loop owns, the same as a user's password, so a unit this
code gains reaches the team that already exists rather than only the
next mesh raised from scratch.
2026-09-25 21:59:09 +02:00
jschoubben bbd1723612 Merge pull request 'route-proxy binds the mesh's own authority for internal names; gitea's internal-API refusal joins its route' (#70) from feat/route-proxy-internal-acme into main 2026-09-25 19:55:32 +00:00
jschoubben c7964b7285 Merge pull request 'gitea: a consumer's user is actually created, and a failed create says why' (#71) from fix/gitea-provisioner-user-create into main 2026-09-25 19:54:08 +00:00
jschoubben 02ccf31a71 gitea: a consumer's user is actually created, and a failed create says why
The builder's package-registry grant — the first this provider ever
received — retried for a day saying only that an edit 404'd. Two faults
under it: the API refuses an email without a dotted domain, so
`@localhost` failed validation at create (the CLI that made mesh-admin
accepts it, which is why the admin exists and no consumer did); and
ensureUser read that 422 as 'already exists' and went on to edit a user
that was never made, burying the create's own message. The address is now
gitea's own hidden-address shape, and the edit path is taken only for a
user that is actually there.
2026-09-25 21:42:38 +02:00
jschoubben c2353fc0a6 gitea: the internal-API refusal is part of the route, not a file beside the proxy
The 2026-09-12 incident response blocked /api/internal by hand in the
predecessor's dynamic directory, with a note that its durable home is the
mesh's routing. A route carries the policy applied to a request (ADR
0108), so the refusal now travels with the grant: route-proxy enforces
it on both the public name and the internal alias the moment it serves
this route, and the adapter skips it aloud (no port, nothing to write)
while the predecessor's own file still stands. The hand-authored file
retires with the proxy it configures.
2026-09-25 20:51:44 +02:00
jochen 45dd036623 The npm registry is a seat gitea holds, and gitea holds the git seat a build's source can live on
Implements novox/hq ADR 0109, 0110 and 0111 in the catalogue.

package-registry becomes npm-package-registry throughout (ADR 0109): gitea provides and serves it,
verdaccio provides it, the builder requires, binds and receives its secret under it. gitea's
contributions file is grants/npm.json, so a second ecosystem's file has an obvious name beside it.

gitea claims two mesh seats (ADR 0110): npm-package-registry, which it delivers, and git, which it
now provides with what a clone URL is composed from — http on the forge's web port (ADR 0111).
verdaccio provides npm-package-registry and claims nothing: it is the second provider the seat
exists to make harmless, since a consumer now resolves to the seat's holder without a pin.

No cargo or PyPI provision is added; ADR 0109 defers that. git mints no credential, so gitea's
provisioner registers nothing for it — the mesh's own repositories are public, and a clone
credential is undecided (ADR 0111).

The provisioner still reads where its contributions land from $MESH_RECEIVES, and names no path
itself. One variable carries one path, so a second registration in this module would need the mesh
to say where each provision's file is; that is not possible yet and is not faked here.

Verified: the controller's tests read this catalogue — every claim is a seat in the set, the forge
holds both seats and serves what a clone URL needs, the builder requires what the npm seat delivers
— and pass. Not verified here: a TypeScript build of gitea, whose dependencies resolve from the
private registry.
2026-09-25 20:48:10 +02:00
jschoubben 962cba7c04 route-proxy: internal names are certified by the mesh's own authority
Two name spaces, two authorities (08-connectivity §2): a public name is
certified by a public CA, an internal one by the mesh's own. step-ca now
offers that second seat as internal-acme-ca beside its existing acme-ca,
and route-proxy requires both — the server dispatches by which authority
may certify the name at all, so an .internal alias stops being plain-HTTP
only without ever asking a public CA for a name it cannot validate.
2026-09-25 20:36:41 +02:00
jschoubben bfe99f8c78 Merge pull request 'minio: declare the route contribution it has always needed, scoped from PR #58' (#68) from fix/minio-declares-its-real-route-contribution into main 2026-09-25 16:19:39 +00:00
jschoubben f0aa9e5fed minio: declare the route contribution it has always needed, scoped from PR #58
files-api.novox.be and files.novox.be worked earlier tonight from route-
adapter-generated files, but minio's module.json on main never actually
carried a route requirement — that capability has been sitting in PR #58
the whole time, bundled with an unrelated network rename and console
redirect URL that need their own calmer review. This is just the two
routes: requires: route, contributes.route.api/.console (ContributesMany,
proven working via mesh-controller #55/#57), and the console port (9001)
actually published and declared in listens.

Found assigning route-proxy for the first time tonight: its own routes
file, generated the identical way route-adapter's always was, had four
hostnames in it instead of six — nothing served files-api/files at all,
which would have been a real, silent outage the moment Traefik stopped.
2026-09-25 18:19:27 +02:00
jschoubben cb38bf08a6 Merge pull request 'route-proxy: the trust step skips the fetch when the CA names no roots to get' (#67) from fix/route-proxy-trust-skips-when-there-is-nothing-to-fetch into main 2026-09-25 16:15:09 +00:00
jschoubben 95a0a5672c route-proxy: the trust step skips the fetch when the CA names no roots to get
Composed ACME_ROOTS unconditionally from ${bound:acme-ca:roots} even when
that field is empty — public-acme's own case, where an empty roots means
'the system trust store', not 'fetch from the bare authority host'. The
run-once step wget'd https://acme-v02.api.letsencrypt.org:443 (host, no
path) for two minutes every apply and failed, blocking every resource
after it — found live tonight, assigning route-proxy for the first time.

Carries the raw, uncomposed roots value alongside the composed URL
(ACME_ROOTS_PATH) so the step can tell 'nothing to fetch' apart from 'the
authority didn't answer' — a distinction the composed URL alone cannot
make. Empty copies the image's own system CA bundle to /ca/root.crt
instead of fetching one, so ACME_CA_BUNDLE stays the one path it has
always been rather than needing to become conditional itself.
2026-09-25 18:14:58 +02:00
jschoubben d243b56942 Merge pull request 'route-proxy: the server container resolves its own built artifact' (#66) from fix/route-proxy-server-uses-its-real-artifact into main 2026-09-25 16:09:19 +00:00
jschoubben 4c5e69903f route-proxy: the server container resolves its own built artifact
Was still pinned to the scaffold's placeholder digest (mesh-route-
proxy@sha256:0000...0000) even after the build+context work landed —
never caught because nothing had assigned route-proxy before tonight.
artifact: server, matching trust's own reference a few lines up and
every other built module in the catalogue.
2026-09-25 18:09:09 +02:00
jschoubben 547937034f Merge pull request 'public-acme: the roots field is named roots, not root' (#65) from fix/public-acme-field-name-matches-what-route-proxy-reads into main 2026-09-25 16:08:39 +00:00
jschoubben 3147fac08b public-acme: the roots field is named roots, not root
step-ca (the other acme-ca provider) already spells it correctly; route-
proxy's own template reads ${bound:acme-ca:roots}. Found live, assigning
public-acme for the first time tonight: the mesh refused the push outright
rather than composing a broken binding — 'route-proxy asks its acme-ca
binding for roots, and what answers it says ... root'. Empty stays empty:
a public CA's root is the system trust store already, per route-proxy's
own design (an empty ACME_CA_BUNDLE means exactly that).
2026-09-25 18:08:28 +02:00
jschoubben 7a96287d99 Merge pull request 'route-proxy: declare the build context its own Dockerfile has always needed' (#64) from feat/route-proxy-declares-its-real-context into main 2026-09-25 15:58:20 +00:00
jschoubben dd5b973e66 route-proxy: declare the build context its own Dockerfile has always needed
The Dockerfile's own comment already said it — 'the build context is the
mesh-controller repository root' — but nothing in the manifest actually
said so to the mesh, so every build attempt used mesh-catalog's own
directory instead and failed with 'stat go.mod: file does not exist'.
Never caught before because route-proxy has never been assigned anywhere.
Uses the context mechanism just added (mesh-controller#62), proven
working tonight on builder's own self-build.
2026-09-25 17:58:10 +02:00
jschoubben 72b1413497 Merge pull request 'builder is a real built module now, not handed over' (#63) from feat/builder-is-a-real-built-module into main 2026-09-25 15:54:58 +00:00
jschoubben 8369fe22b8 builder is a real built module now, not handed over
Its own image ('mesh-builder@sha256:0000...0000', later manually pinned to
a real digest tonight when the placeholder blocked a push) was never
produced by anything the mesh tracks — cmd/mesh-builder lives in
mesh-controller's own repository, and nothing declared how to build an
image from it. Uses the same context mechanism route-proxy does (mesh-
controller#62): the Dockerfile compiles ./cmd/mesh-builder from a clone of
mesh-controller's repository, not a vendored copy.

Unlike mesh-controller's own FROM scratch (ADR 0006 — nothing to audit
but one binary), the build machine's whole job is shelling out to git and
docker, so its runtime is Alpine with both installed from the base's own
packages, not fetched on their own.

Bootstrapped live tonight: a manual build got the new image running long
enough to build itself properly through the pipeline it had just gained,
and mesh-controller itself needed the same upgrade first (it parses
manifests too, and rejected the new context field with the old binary) —
genesis's own kind of ordering problem, solved by hand exactly once.
2026-09-25 17:54:46 +02:00
jschoubben 19e0a8469d Merge pull request 'minio: install mc in the runtime image, declare its region' (#62) from fix/minio-provisioner-missing-mc-cli into main 2026-09-25 15:25:18 +00:00
jschoubben 271911753f Merge pull request 'builder: consume package-registry as a real mesh grant, not a hand-faked one' (#61) from fix/builder-real-package-registry-grant into main 2026-09-25 15:25:04 +00:00
jschoubben 4e2dd0adae Merge pull request 'gitea: listens its real internal ssh port (22), not 2222' (#60) from fix/gitea-ssh-port-manifest-mismatch into main 2026-09-25 15:24:49 +00:00
jschoubben 1b02dc1f66 builder: qualify its own image with the registry host
Bare mesh-builder@sha256:... is only resolvable for a module with a build
section — the mesh's own build step rewrites the reference to a real
registry path as part of resolving build.artifacts. builder is handed
over, not built, so nothing ever rewrites it: pushed as written, docker
read it literally and tried Docker Hub. Took the live node's mesh-builder
down for the length of one push-and-fix (docker: pull access denied for
mesh-builder, repository does not exist). novox.internal:5100, not the
literal external IP docker inspect showed live, for the same reason
addresses generally don't get hardcoded in this catalogue.
2026-09-25 17:10:56 +02:00
jschoubben 755b0a5599 builder: consume package-registry as a real mesh grant, not a hand-faked one
The 'package-binding' resource was a hardcoded JSON fragment standing in
for a real grant — {"provision": "package-registry", "from": "gitea",
"at": "127.0.0.1", ...} written as if it were mesh-resolved, when nothing
resolved it. Declares requires: package-registry properly instead, with
binds/secrets pointing at the same file paths the resource used to
manually author, so the mesh mints the grant and writes it there.

npm-password renamed to package-registry.secret: it's gitea's generic
user+password, not npm-specific — the same credential works for basic
auth against cargo/PyPI/Go package endpoints too, once gitea's manifest
grows them (novox/hq ADR 0109).

Known gap, not fixed here (novox/hq issue 117): this makes builder
correct for the steady state but breaks a genesis bootstrap — gitea's own
image is built by builder, so builder cannot yet hold this grant the
first time either has to exist. Filed rather than silently accepted.
2026-09-25 17:08:12 +02:00
jschoubben ae6dfea2c9 gitea: listens its real internal ssh port (22), not 2222
2222 never meant anything — no container of gitea's publishes it, the
software never listens on it, and nobody could say where it came from.
The container's real internal sshd is 22 (gitea's own default, unmodified
— every other module's listens.port already means the container's real
internal port, this one didn't). ports now declares 222:22 directly: 222
is the real, fixed, always-known public git-ssh port, so it needs no
per-node setting to reach — unlike an arbitrary auto-assigned port, this
one has external consumers who already know the number.
2026-09-25 17:07:58 +02:00
jschoubben b3de7e7944 minio: declare region eu-west, don't leave it as live-only drift
serves.s3-bucket.region still said us-east-1 (PR #58 already fixed this,
unmerged) while the live mesh-minio sidecar had MESH_MINIO_REGION=eu-west
set out-of-band, not in the manifest at all — and the actual minio server
had no region configured whatsoever (mc admin config get region: empty),
apparently lost across a container recreation since nothing declared it.
Every s3-bucket consumer binding ${bound:s3-bucket:region} was reading
the stale us-east-1 declaration regardless of what was actually live.

Declares MINIO_REGION on the server container and MESH_MINIO_REGION on
the sidecar, matching the static serves declaration, so this is mesh-
managed and durable rather than a manual mc admin config or docker env
override that the next recreation silently drops.
2026-09-25 15:10:30 +02:00
jschoubben 107090310d nextcloud: point S3 config at the bucket the mesh actually provisioned
OBJECTSTORE_S3_BUCKET=nextcloud was a leftover from before the module
existed — that bucket was created by hand during tonight's earlier HAL
credential stopgap. The mesh's own minio provisioner derives its own
bucket name from the consumer's access-key identity (bucketFor(as) in
minio/client.ts) rather than honouring contributes.s3-bucket.bucket — by
design, so teardown can recompute the name with nothing persisted — and
minted mesh-novox-ncloud, a different bucket. mesh_novox_ncloud's scoped
policy only covers that bucket, so every S3 write 403'd with AccessDenied
trying to touch the old one. Pointed both the request hint and the real
env var at the bucket that's actually there.
2026-09-25 14:42:57 +02:00
jschoubben 5d13a5078c minio: install mc in the runtime image — the provisioner needs it to run
mesh-minio's s3-bucket provisioner shells out to mc to create buckets and
service accounts on the live server, but mc was never in this module's own
runtime image, only in minio's own. It's been silently retrying 'spawn mc
ENOENT' forever, so every s3-bucket grant reached the control-plane layer
(store.json, sealed secret) without the credential ever actually existing
on minio — nextcloud's live instance just hit this as InvalidAccessKeyId
on a real user session.

Copies mc from minio's own image (docker.io/pgsty/minio, already pinned
and pulled as this module's server container) rather than introducing a
new base — mc there is a working, already-verified binary. /usr/bin/mc is
a symlink to mcli; both are copied so it resolves.
2026-09-25 14:37:23 +02:00
jschoubben 61eca75315 nextcloud: pull the S3 region from the binding, not a hardcoded literal
the s3-bucket binding already carries serves.region (same mechanism as
at/port); ${bound:s3-bucket:region} tracks whatever minio is actually
configured with instead of a copy that can drift
2026-09-25 14:15:40 +02:00
jschoubben df0e9017e0 nextcloud: set OBJECTSTORE_S3_REGION to match minio's actual region
minio runs with MINIO_REGION=eu-west; nextcloud's S3 config never set a
region, so every object write (avatars, file writes) failed signature
validation with AuthorizationHeaderMalformed, surfacing as Internal Server
Error on real page loads
2026-09-25 14:12:05 +02:00
jschoubben b60dd5e3e3 nextcloud: use a module-owned mesh-admin account instead of 'admin'
the migrated data has no literal 'admin' account — HAL's real admin login
is a personal account (jochens), not a generic one. Resetting that would
touch a real user's own credential, so the module gets its own dedicated
admin-group service account instead, same pattern as the minio per-module
service accounts. mesh-admin was created once by hand on novox to match
this manifest for the already-migrated data; a genuinely fresh install
seeds it automatically via NEXTCLOUD_ADMIN_USER/NEXTCLOUD_ADMIN_PASSWORD.
2026-09-25 13:51:40 +02:00
jschoubben b865978e19 nextcloud: use a named build stage for docker:cli, not ARG-in-COPY-from
the legacy (non-BuildKit) docker build this host runs doesn't expand ARGs
inside COPY --from — only FROM. Give it its own named stage instead
2026-09-25 13:49:02 +02:00
jschoubben cdbc850c52 nextcloud: declare docker:cli as a pinned build.on base
build refused to reach docker:cli implicitly (novox/hq ADR 0097); pin it by
digest and thread it through as DOCKER_CLI, redeclared in the final stage
since args declared before the first FROM don't carry past it
2026-09-25 13:48:00 +02:00
jschoubben e2b3723dd9 nextcloud: fix hardcoded sidecar port and missing docker CLI in runtime image
- MESH_NEXTCLOUD_URL hardcoded :80 instead of the mesh-assigned ${port:80}
- sidecar's occ() shells to docker exec but the docker CLI binary was never
  present in the runtime image, only the mounted socket
2026-09-25 13:47:03 +02:00
jschoubben 5f74c41a3e nextcloud: give the admin password a real _FILE variant, not an env-file
The mesh's own check caught it: an env-file-loaded secret still reaches
the process environment, readable via docker inspect and /proc (hq
04-ISSUES/041) -- the same class of exposure the file-based delivery
exists to avoid. Added MESH_NEXTCLOUD_ADMIN_PASSWORD_FILE support to the
client, matching the pattern the minio client already uses, and mounted
the sealed admin secret directly rather than writing it into an env-file.
2026-09-25 13:42:21 +02:00
jschoubben 4329ca9392 nextcloud: deliver the admin password to the sidecar too
The sidecar's own client needs MESH_NEXTCLOUD_ADMIN_PASSWORD to list
shares over the OCS API, but the runtime container's env/volumes never
carried it -- only the server container did. Delivered the same way every
other sealed value in this manifest already is: a generated env-file with
the ${secret:admin} substitution, not a raw value in the container's env.
2026-09-25 13:41:17 +02:00
jschoubben 066254e5c4 nextcloud: pin an image whose PHP matches this Nextcloud version
The pinned digest resolved to a PHP 8.5.10 image; Nextcloud 30 refuses to
run above PHP 8.4. Repinned to the current digest for the nextcloud:30 tag
(matches HAL's own NEXTCLOUD_VERSION), which carries PHP 8.3.28 -- the
same version the data being migrated was actually running under.
2026-09-25 13:39:16 +02:00
jschoubben 440e3e446e minio: set the region to eu-west, matching where this mesh actually runs
Left at MinIO's us-east-1 default. Novox is hosted in Germany, the team
is in Belgium -- eu-west is correct, and matters beyond labeling: it's
part of the SigV4 signature, so a client using the wrong region fails
auth even with valid credentials. Set on the server (MINIO_REGION),
the served provision value, and the runtime sidecar's own client.
2026-09-25 10:45:01 +02:00
jschoubben 20df40c949 minio: revert to single-node after measuring the real cost of sharding
The 4-node/8-drive erasure-coded cluster matched HAL's topology faithfully,
but real throughput testing against both showed why that costs more than
it's worth here: every write on the sharded cluster fans out across 4
processes over the internal network with erasure-coding overhead, capping
safe throughput around 1.3-2 MiB/s and breaking outright above ~256
concurrent transfers (IncompleteBody errors, confirmed via a controlled
512x test). The identical copy against a single-node instance sustained
23+ MiB/s at the same concurrency with zero errors — over 10x faster,
verified side-by-side, not assumed.

Trades away erasure-coded redundancy (no single-drive fault tolerance) for
that throughput. Deliberate, and reversible if it turns out to matter later
-- the data itself is migrated over the S3 API either way, so the storage
topology underneath isn't locked in by anything upstream of it.
2026-09-24 23:07:02 +02:00
jschoubben 6a6dd4a7dc minio: name the network minio-net, not minio
Collided with the LB container's own name. docker inspect minio resolved
to the network instead of the (not-yet-created) container, and mesh-host's
existence check crashed on the mismatched shape rather than reporting
absence -- a real mesh-host bug (fixed separately, mesh-host#25), but this
sidesteps it here without waiting on a host-level binary update.
2026-09-24 20:31:20 +02:00
jschoubben 973d80aaa2 minio: publish both public routes now that a module can answer route twice
files-api.novox.be (port 9000, the S3 data API) and files.novox.be (port
9001, the console) — same two names HAL routes today, via nginx's own
upstream split. Needed mesh-controller#55 (a module answering one
requirement several times) to exist first; it's merged and deployed.
2026-09-24 18:45:52 +02:00
jschoubben 945390e59a minio: run the real 4-node/8-drive erasure-coded cluster, not a single container
The single standalone instance from the first pass didn't match HAL's actual
topology: HAL runs minio1-4, two drives each, behind an nginx load balancer
on 9000 (S3) and 9001 (console). This rewrite mirrors that exactly — same
node count, same erasure-coding command, same LB config — so the migration
is a real like-for-like move, not a simplification.

Only the two images that had to change did: the minio server (dead upstream,
already fixed in the prior commit) and nginx (1.19.2-alpine is long EOL;
repinned to current stable-alpine by digest). Data still lands on a fresh,
empty, mesh-owned path, never HAL's live drives.

The OIDC-wait entrypoint wrapper HAL used is dropped: it's a no-op when
MINIO_IDENTITY_OPENID_CONFIG_URL is unset (it always is here — no OIDC
integration was ever wired to minio itself), and this catalogue has no
container resource field for overriding a container's entrypoint anyway —
every converted module relies on the image's own entrypoint plus args,
which is exactly what the original single-node version already did.
2026-09-24 18:31:38 +02:00
jschoubben 49c5903861 Merge pull request 'minio: repin to pgsty's fork, build the runtime sidecar, move data off HAL's live drive' (#57) from fix/minio-repin-and-move-off-hal-data into main 2026-09-24 16:15:09 +00:00
jschoubben 35e8a3fe19 minio: repin to pgsty's fork, build the runtime sidecar, move data off HAL's live drive
minio/minio and minio/mc were pulled from all public registries on 2026-09-11;
pgsty's fork is the working replacement (hq issue 113). The runtime sidecar
had no Dockerfile and no build section at all — added, following the same
tsc-over-client/tools/provisioner shape every other converted module uses.

The data resource pointed straight at /services/minio/data/data1-1, one of
HAL's live 8-drive erasure-coded array — starting this module would have
written into production storage the mesh doesn't own. Moved to a fresh,
empty, mesh-owned directory; the actual data migration happens over the S3
API (rclone), not by sharing a disk path.
2026-09-24 17:55:49 +02:00
jschoubben ed0f4602a6 keycloak: use Hostname v2's actual config shape, not v1's deprecated flags
The previous commit on this branch used KC_PROXY=edge and
KC_HOSTNAME_STRICT_HTTPS=true, carried over from HAL's config -- but
HAL ran an older Keycloak using the v1 hostname provider. This image
(26.0.8) defaults to Hostname v2, which warned 'options [proxy,
hostname-strict-https] are still in use, please review your
configuration' and kept generating http:// URLs regardless -- verified
against /realms/Novox/.well-known/openid-configuration directly, not
just the login button, after the first fix deployed.

v2's actual shape (keycloak.org/server/hostname): KC_HOSTNAME is a full
URL, not a bare hostname -- the scheme in the URL is what tells Keycloak
to generate https, not a separate strict-https flag. KC_PROXY_HEADERS
replaces KC_PROXY: xforwarded to trust traefik's X-Forwarded-* headers,
which it sends by default.
2026-09-24 17:27:04 +02:00
jschoubben 13d0361640 keycloak: carry over HAL's hostname/proxy settings, dropped during conversion
Reported: files.novox.be's login button redirects to http://keycloak.novox.be,
not https. HAL's original config (/services/keycloak/docker-compose.yml) set
three settings the mesh's manifest never carried over:

  KC_HOSTNAME: keycloak.novox.be
  KC_HOSTNAME_STRICT_HTTPS: true
  KC_PROXY: edge

Without KC_PROXY: edge, Keycloak has no way to know it sits behind a
TLS-terminating reverse proxy (traefik) -- it generates URLs from what it
directly sees, which is plain HTTP from traefik's backend connection. Same
pattern as the named-volume conversion: the shape was rebuilt from general
knowledge of what a keycloak container needs, not from what this
installation's own working config actually had.
2026-09-24 17:25:16 +02:00
jschoubben 61eb201f8a postgres: declare the data directory's real owner; keycloak: use the port template
postgres: mesh-store's data directory has always had split ownership --
everything inside pgdata/ is owned by UID 999 (the pgvector image's real
runtime user), while only the top-level mount point happened to be 70:70.
Invisible while the directory's mode was 1777 (world-accessible, from the
named volume this replaced); broke the moment mode: 0700 was enforced,
locking out the actual owning process. mesh-store crash-looped on
Permission denied twice before this was found -- once at container
creation, once mid-session on a checkpoint, after ownership looked correct
by every check that didn't look inside pgdata/ specifically.

keycloak: MESH_KEYCLOAK_URL was hardcoded to :8080, but the module's own
port override (settings set keycloak {ports:{8080:28080}} on novox) means
the real published port is 28080. Same bug class as the postgres
connection-string fix earlier tonight -- now using the mesh's own
 template instead, which is exactly the mechanism
internal/catalogue/port_into.go describes for a sidecar dialling its own
server over the machine's loopback.
2026-09-24 16:39:11 +02:00
jschoubben 5c3781d8c2 Merge pull request '4 modules: persistent data is a directory bind, never a named volume' (#54) from fix/persistent-data-is-a-directory-bind-not-a-volume into main 2026-09-24 14:24:15 +00:00
jschoubben 6a3e0ab9d3 4 modules: persistent data is a directory bind, never a named volume
hq ADR 0107 / issue 115. HAL's own postgres and lavinmq both used a
directory bind (./db-data, ./data) for exactly this data -- the mesh's
adoption of them, three weeks ago, switched to a named Docker volume
instead, and searxng/distribution followed the same pattern since.

A named volume survives ordinary container recreation, same as a
directory bind -- that was never the problem. The problem is everything
else: docker rm -fv, docker volume rm, and docker system prune --volumes
all target it (one flag away from the docker rm -f this migration already
uses routinely); it is invisible to every tool this migration has used all
night (ls, find, grep across /var/lib, /services); and nothing outside
Docker's own volume machinery can back it up or notice it growing.

mesh-store carries the sharpest version: every database migrated tonight,
including keycloak's, live inside it.

Data already copied and verified on novox before this merges:
- mesh-registry-data -> /var/lib/mesh-registry (11G, diff -rq clean)
- mesh-broker-data -> /var/lib/mesh-broker (37M, cp -a)
- mesh-broker-tls -> /var/lib/mesh-broker-tls (12K, cp -a)
- mesh-store-data -> /var/lib/mesh-store (1.7G) -- mesh-store stopped
  cleanly first, so the final copy is crash-consistent, not a live-file
  copy of a running postgres; diff -rq clean after.
- searxng/valkey: not yet assigned anywhere, manifest-only fix, nothing
  to copy.

Old named volumes left in place, not deleted, as the rollback path.
2026-09-24 16:12:03 +02:00
jschoubben 3e8ecfa4eb Merge pull request '11 modules: a container's ports say the software's port, not the machine's' (#53) from fix/091-ports-machine-side-is-the-mesh-to-assign into main 2026-09-24 13:53:34 +00:00
jschoubben 12f885ffc5 11 modules: a container's ports say the software's port, not the machine's
hq issue 091, measured 2026-09-22: 14 of 46 modules with containers fix the
machine side of a published port in their own manifest -- a fact about one
machine (which port HAL happened to publish it on) written into a
definition meant for any node. ADR 0038 already says the mesh assigns the
machine side and a module says only what it needs; the machinery already
does it (internal/inventory/ports.go's PortFor, declaration.go's
publishedOn rewrites a bare port automatically). These 14 just never
complied.

Fixed 11 of them -- stripped to the bare software port, letting assignment
take over: de-spiegel, gitea, hello-web (the demo module), mailu, mssql,
n8n, novox.be, only-office, photos, photos-eef, photos-filip.

Left alone, the two defensible kinds the issue names: postgres/lavinmq/
distribution (foundation, genesis-rewritten per ADR 0100 -- the number in
the manifest is a default, not a claim) and unifi (protocol/
device-discovery fixes the number; nothing else can find the controller).

gitea was a live one, not just a tidiness fix: its manifest said 2222:22,
but the actual adopted, running container is on 222. Harmless while held,
but the next 'take' would have recreated it on the wrong port and broken
SSH git access. Recorded 222 as novox's own setting for it (settings set
gitea -node novox) so that doesn't happen.
2026-09-24 15:45:58 +02:00
jschoubben 7551a65579 Merge pull request 'postgres: the provisioner connects to mesh-store's actual port, not a hardcoded 5432' (#52) from fix/postgres-provisioner-connects-to-the-actual-store-port into main 2026-09-24 13:38:45 +00:00
jschoubben 135101ce6e postgres: fix the port override properly — a twin variable, not a raw substitution
The first commit on this branch embedded ${seat:mesh-store:5432} directly
inside MESH_PROVISION_POSTGRES's URL. That placeholder resolves to an empty
string whenever mesh-store is on its own default port (novox/hq
internal/catalogue/seat_into.go: 'the mesh raised on the catalogue's own
ports never gives them a setting at all') -- which produces a malformed
connection string (host:/postgres) on exactly the common case, a fresh,
non-adopted mesh. It only worked here because novox's mesh-store happens to
be adopted at a non-default port.

mesh-controller's own manifest already has the right shape for this --
internal/envfile/port.go's NAME / NAME_PORT twin, composed by
mesh-controller's Placed(): the base value keeps its own port; a separate
_PORT variable carries the override, spliced in only when it says
something, and left alone -- not a fault -- when it's an unfilled
placeholder (a manifest ahead of the running controller).

Reverted the manifest to its original base value, added
MESH_PROVISION_POSTGRES_PORT as the twin, and taught client.ts (the only
consumer -- both tools/ and provisioner/ import it) the same precedence
envfile.Placed uses. Checked: no other file in the module reads
MESH_PROVISION_POSTGRES directly.
2026-09-24 15:27:16 +02:00
jschoubben 088022d63b postgres: the provisioner connects to mesh-store's actual port, not a hardcoded 5432
MESH_PROVISION_POSTGRES was a literal connection string naming port 5432 --
correct only when mesh-store happens to run on the mesh's own default. On
novox, mesh-store was adopted in place at HAL's original port (6852), and
the provisioner has been retrying-and-failing against 127.0.0.1:5432 ever
since, for every consumer including ones that already exist (gitea, umami,
mesh-catalog), not just a new grant.

Fixed with the same ${seat:mesh-store:5432} template mesh-controller's own
manifest already uses for the identical connection. No other module needed
this fix checked -- lavinmq's MESH_PROVISION_LAVINMQ already used
127.0.0.1:15672 unconditionally, but the broker's management port is fixed
by the module itself (127.0.0.1:15672 in lavinmq/module.json's own ports),
not by adoption, so it isn't the same bug.
2026-09-24 15:24:07 +02:00
jschoubben 415ad7a168 Merge pull request 'gitea: the token needs read:user, not just write:repository and write:issue' (#51) from fix/gitea-user-repos-scope into main 2026-09-24 09:49:36 +00:00
jschoubben aaf341a2d8 gitea: the token needs read:user, not just write:repository and write:issue
Deployed #49 and the watcher immediately broke: GET /user/repos answered 403,
'required=[read:user]' — confirmed live against the running forge (1.27.3).
That route sits under gitea's user scope category despite listing
repositories, not repository as assumed.

Also gives the fake forge real scope enforcement on /user/repos, which is
why the original PR's test suite didn't catch this: it only checked the
token's value was valid, never that it carried the required scope.
2026-09-24 11:43:46 +02:00
jschoubben 8a458aa5c4 Merge pull request 'The forge mints its own API token with the admin account the vault delivers' (#49) from fix/gitea-mints-its-token into main 2026-09-24 08:41:13 +00:00
jschoubben d036b43bfc Merge main into fix/gitea-mints-its-token to pick up the resolver conversion and artifact-store seat fix 2026-09-24 10:40:57 +02:00
jschoubben 5a94b78891 Merge pull request 'Convert the resolver from the module it replaces: forward, answer at 127.0.0.1, point the runtime at it' (#50) from convert/dnsmasq-from-hal into main 2026-09-23 23:13:08 +00:00
jschoubben ed094031fd The forge mints its own API token with the admin account the vault delivers
The runtime beside the forge served 0 tools: GiteaClient.fromEnv required a token
(settings or MESH_GITEA_TOKEN), nobody had one to give — the mesh raised the forge —
and putting one in settings would store a secret in plaintext in the inventory. So
the fifteen tools registered nothing and the watcher logged "not watching".

What the mesh does deliver is the admin account: a login the manifest names and a
password the vault minted and the host unsealed into a file (ADR 0086). That is
enough to mint a token, so the module does (hq issue 100, the forge's tools):
POST /users/{admin}/tokens over basic auth, scoped to write:repository and
write:issue — the least the tools and the repo watcher need — kept at 0600 in the
module's own state (/var/lib/mesh/gitea/state, a new directory resource the runtime
mounts writable), read back on the next start, and minted afresh when the forge
answers 401 to it or the kept file is gone. A forge whose data came from the
predecessor has no mesh-admin: that is reported in plain words on every poll until
it clears, once per reason, not crash-looped. A configured token still wins and is
never minted over.

The mint happens on the first call, not at registration: a contributor is
synchronous, and a forge not yet answering must not keep the runtime from serving.
One source per kept file in a process, or the watcher and the tools would each
renew on a 401 and drop the other's token by name.
2026-09-23 23:48:36 +02:00
92 changed files with 2635 additions and 758 deletions
+25
View File
@@ -0,0 +1,25 @@
ARG GO_BASE
ARG ALPINE_BASE
# builder's own image: the build machine itself, compiled into a container.
#
# **The source is not vendored here.** builder's actual code — cmd/mesh-builder, internal/builder,
# internal/catalogue — lives in the mesh-controller repository, the same control plane it is one
# half of. This module ships the packaging, not a second copy of the source, so the build context
# is the mesh-controller repository root (declared under build.artifacts[].context), and this
# Dockerfile compiles ./cmd/mesh-builder from it — the same shape route-proxy already uses for the
# same reason.
FROM ${GO_BASE} AS build
WORKDIR /src
COPY go.mod go.sum ./
RUN go mod download
COPY . .
RUN CGO_ENABLED=0 GOOS=linux go build -trimpath -o /mesh-builder ./cmd/mesh-builder
# Unlike mesh-controller's own FROM scratch (ADR 0006: nothing to audit but one binary), the build
# machine's whole job is shelling out to git and docker — it needs a real userland to do that in,
# not a second copy of either tool vendored into this image. apk installs both from the base's own
# packages, not fetched on its own at build time.
FROM ${ALPINE_BASE}
RUN apk add --no-cache docker-cli git
COPY --from=build /mesh-builder /usr/local/bin/mesh-builder
ENTRYPOINT ["/usr/local/bin/mesh-builder"]
+38 -25
View File
@@ -6,19 +6,25 @@
],
"claims": [
{
"name": "the-build-machine",
"scope": "node"
"name": "mesh-build-machine",
"scope": "mesh"
}
],
"requires": [
"artifact-store"
"artifact-store",
"npm-package-registry"
],
"binds": {
"npm-package-registry": "/var/lib/mesh/builder/package-registry.json"
},
"secrets": {
"npm-package-registry": "/var/lib/mesh/builder/package-registry.secret"
},
"emits": [
"module.builder.built"
],
"own-secrets": {
"broker": "/var/lib/mesh/builder/broker",
"npm-password": "/var/lib/mesh/builder/npm-password"
"broker": "/var/lib/mesh/builder/broker"
},
"resources": [
{
@@ -38,27 +44,13 @@
"type": "file",
"path": "/var/lib/mesh/builder/builder.env",
"mode": "0600",
"content": "MESH_BROKER_FILE=/run/mesh/broker\nMESH_NODE=${machine:name}\nMESH_REGISTRY=${bound:artifact-store:at}:${bound:artifact-store:port}\nMESH_PACKAGE_BINDING=/run/mesh/package-registry.json\nMESH_NPM_TOKEN_FILE=/run/mesh/npm-password\nMESH_WORKSPACE=/var/lib/builder/workspace\n"
},
{
"id": "package-binding",
"type": "file",
"path": "/var/lib/mesh/builder/package-registry.json",
"mode": "0600",
"merge": "json",
"protected": [
"provision",
"from",
"at",
"as"
],
"content": "{\"provision\": \"package-registry\", \"from\": \"gitea\", \"at\": \"127.0.0.1\", \"as\": \"mesh-builder\", \"serves\": {\"scheme\": \"http\", \"port\": 3000, \"npm-path\": \"/api/packages/novox/npm/\"}}\n"
"content": "MESH_BROKER_FILE=/run/mesh/broker\nMESH_NODE=${machine:name}\nMESH_REGISTRY=${bound:artifact-store:at}:${bound:artifact-store:port}\nMESH_PACKAGE_BINDING=/run/mesh/package-registry.json\nMESH_NPM_TOKEN_FILE=/run/mesh/package-registry.secret\nMESH_WORKSPACE=/var/lib/builder/workspace\n"
},
{
"id": "server",
"type": "container",
"name": "mesh-builder",
"image": "mesh-builder@sha256:0000000000000000000000000000000000000000000000000000000000000000",
"artifact": "server",
"env-file": [
"/var/lib/mesh/builder/builder.env"
],
@@ -68,11 +60,32 @@
"/var/run/docker.sock:/var/run/docker.sock"
],
"restart-on": [
"builder-env",
"package-binding",
"needs-npm-password"
"builder-env"
],
"network": "host"
}
]
],
"build": {
"artifacts": [
{
"name": "server",
"kind": "image",
"from": "Dockerfile",
"context": {
"repository": "https://git.novox.be/novox/mesh-controller.git",
"ref": "main"
}
}
],
"on": [
{
"arg": "GO_BASE",
"image": "golang@sha256:1ae0735f00daffa3aaf1363a5184c0d2dc55c78e3db4ec70241cdac97bf84b59"
},
{
"arg": "ALPINE_BASE",
"image": "alpine@sha256:d9e853e87e55526f6b2917df91a2115c36dd7c696a35be12163d44e6e2a4b6bc"
}
]
}
}
+1 -1
View File
@@ -58,7 +58,7 @@
"/var/lib/de-spiegel/server.env"
],
"ports": [
"35621:35621"
"35621"
],
"secrets-in-environment": "the application's own code reads SMTP_AUTH_USER/PASS from the environment (de-spiegel server/index.js); converting is that repository's change"
}
+43
View File
@@ -0,0 +1,43 @@
# dhcpcd
The uplink seat's module for a machine whose own network is dhcpcd's (novox/hq ADR 0117). It
asks two things of dhcpcd, and nothing else: leave the resolver file to the mesh, and leave the
private network's interface alone. It never declares an interface, an address, a route, a
wireless network or its credentials — the link dhcpcd keeps is the only channel the mesh reaches
the machine over.
## What it writes
Two lines into `/etc/dhcpcd.conf`, as the mesh's marked region (`into: block`) — dhcpcd reads no
drop-in directory, so the mesh writes into its one file rather than over it (ADR 0102):
- `nohook resolv.conf` — dhcpcd's resolv.conf hook rewrites `/etc/resolv.conf` on every lease it
takes or renews, which would silently replace the resolver `resolv-conf` names.
- `denyinterfaces mesh0` — dhcpcd never asks for a lease on the private network's interface, and
never takes it down. dhcpcd leaves a point-to-point interface alone by default; this says so
rather than relying on it.
**At the start of the file** (`at: start`). Both are global options, and dhcpcd reads every line
after an `interface` or `ssid` line as that interface's own. A configured machine's file ends in
exactly such a block (the interface, its static address), so appended at the end these two would
quietly apply to one interface only.
## Why it declares no service
dhcpcd is the machine's, not the mesh's. The mesh never starts, stops or enables it: stopping it
drops the address the machine is reached at, and a module unassigned by mistake must not be able
to do that. And there is nothing to reload it with — `dhcpcd.service` reports `CanReload=no`, and
a restart drops the lease. So the two lines take effect at **dhcpcd's next start**.
On an adopted machine that is normally no gap: the predecessor wrote the same `nohook` line, and
it is already in force. **On a machine that was not adopted, it is one:** until dhcpcd next
starts (a reboot, or the operator restarting it in a window of their choosing), a lease renewal
still rewrites `/etc/resolv.conf`, and `resolv-conf` puts it back at the next push. Assign this
module before `resolv-conf` on such a machine, and restart dhcpcd once, by hand, when losing the
link for a moment is acceptable.
## One manager per machine
It claims `the-uplink`: a machine runs one network manager, and assigning a second module that
claims the seat is refused. Assigning this one to a machine whose network is NetworkManager's
installs the package and writes the two lines, and starts nothing.
+30
View File
@@ -0,0 +1,30 @@
{
"module": "dhcpcd",
"version": "1",
"capabilities": [
"package-manager",
"service-manager"
],
"claims": [
{
"name": "node-uplink",
"scope": "node"
}
],
"resources": [
{
"id": "package",
"type": "package",
"package": "dhcpcd"
},
{
"id": "config",
"type": "file",
"path": "/etc/dhcpcd.conf",
"mode": "0644",
"into": "block",
"at": "start",
"content": "# The mesh's two lines (module dhcpcd, novox/hq ADR 0117). Global options, so\n# kept above any interface line; read at dhcpcd's next start.\nnohook resolv.conf\ndenyinterfaces mesh0\n"
}
]
}
+7 -1
View File
@@ -42,6 +42,12 @@
"path": "/var/lib/mesh/registry",
"mode": "0700"
},
{
"id": "registry-data",
"type": "directory",
"path": "/var/lib/mesh-registry",
"mode": "0700"
},
{
"id": "store",
"type": "container",
@@ -51,7 +57,7 @@
"5000:5000"
],
"volumes": [
"mesh-registry-data:/var/lib/registry"
"/var/lib/mesh-registry:/var/lib/registry"
]
}
]
File diff suppressed because one or more lines are too long
+2 -2
View File
@@ -6,7 +6,7 @@
],
"claims": [
{
"name": "the-intrusion-prevention",
"name": "node-intrusion-prevention",
"scope": "node"
}
],
@@ -33,7 +33,7 @@
"type": "file",
"path": "/etc/fail2ban/jail.local",
"mode": "0644",
"content": "[INCLUDES]\n\nbefore = paths-arch.conf\n\n[DEFAULT]\n\nbantime = 10m\nfindtime = 10m\nmaxretry = 5\n\nbanaction = ufw\nbanaction_allports = iptables-allports\n\n[sshd]\nenabled = true\nport = ssh\nlogpath = %(sshd_log)s\nbackend = %(sshd_backend)s\n"
"content": "[INCLUDES]\n\nbefore = paths-arch.conf\n\n[DEFAULT]\n\n# Never act on the machine itself or on a tunnel peer: the mesh's private range is\n# ${machine:mesh-range}, named here rather than written as a value the module cannot\n# know (novox/hq ADR 0112). Without this, fail2ban could ban the mesh's own nodes.\nignoreip = 127.0.0.1/8 ::1 ${machine:mesh-range}\n\nbantime = 10m\nfindtime = 10m\nmaxretry = 5\n\nbanaction = ufw\nbanaction_allports = iptables-allports\n\n[sshd]\nenabled = true\nport = ssh\nlogpath = %(sshd_log)s\nbackend = %(sshd_backend)s\n"
},
{
"id": "jail-sshd",
+1 -1
View File
@@ -23,7 +23,7 @@ COPY . .
# The compiler is invoked by its real path rather than through node_modules/.bin, whose entries are
# symlinks to a launcher that requires its library relatively — resolved away when the base image
# was assembled.
RUN node /app/node_modules/typescript/bin/tsc client.ts index.ts provisioner/index.ts tools/index.ts \
RUN node /app/node_modules/typescript/bin/tsc client.ts token.ts index.ts provisioner/index.ts tools/index.ts \
--module NodeNext --moduleResolution NodeNext --target ES2022 --outDir dist
FROM ${RUNTIME_BASE}
+102 -43
View File
@@ -4,6 +4,7 @@
// does.
import { readFileSync } from "node:fs";
import { ConfiguredToken, MintedToken, type TokenSource } from "./token.js";
/** A repository, trimmed to what the mesh cares about. */
export interface GiteaRepo {
@@ -53,44 +54,60 @@ function meshConfig(file?: string): Record<string, string> {
export class GiteaClient {
readonly baseUrl: string;
private cachedUsername: string | null = null;
private readonly tokens: TokenSource;
constructor(
url: string,
private readonly token: string,
) {
/** A token given as a string is one somebody configured; a source decides for itself (token.ts). */
constructor(url: string, token: string | TokenSource) {
this.baseUrl = url.replace(/\/+$/, "");
this.tokens = typeof token === "string" ? new ConfiguredToken(token) : token;
}
/**
* Build from the module's resolved environment. URL and token come from MESH_GITEA_URL /
* MESH_GITEA_TOKEN (the mesh's own names), falling back to the bare GITEA_* names and, for the
* URL, to the forge's loopback port. A token is required — without one there is no authenticated
* call to make, so this throws rather than hand back a client that fails on first use.
* Build from the module's resolved environment. The URL comes from MESH_GITEA_URL (the mesh's own
* name), falling back to the bare GITEA_URL and to the forge's loopback port. The token, in order:
* one configured in settings or the environment (MESH_GITEA_TOKEN / GITEA_TOKEN), which wins; else
* one the module mints for itself with the admin account the vault delivered and keeps in its own
* state (token.ts; hq issue 100). Throws only when neither is possible, naming what is missing,
* rather than hand back a client that fails on first use.
*/
static fromEnv(env: NodeJS.ProcessEnv = process.env): GiteaClient {
const cfg = meshConfig(env.MESH_GITEA_CONFIG_FILE);
const url = cfg.url ?? env.MESH_GITEA_URL ?? env.GITEA_URL ?? `http://127.0.0.1:${env.GITEA_PORT ?? "3000"}`;
const token = cfg.token ?? env.MESH_GITEA_TOKEN ?? env.GITEA_TOKEN;
if (!token) throw new Error("no Gitea token — set MESH_GITEA_TOKEN");
return new GiteaClient(url, token);
const configured = cfg.token ?? env.MESH_GITEA_TOKEN ?? env.GITEA_TOKEN;
if (configured) return new GiteaClient(url, new ConfiguredToken(configured));
return new GiteaClient(url, MintedToken.fromEnv(url, env));
}
/**
* One authenticated call. A 401 is the forge saying the token is not one it knows — the case
* after the forge's data was restored, or after somebody revoked it — so the source is asked to
* renew once and the call is repeated with the new token. A configured token has nothing to renew
* with, and its source says so.
*/
private async request<T = unknown>(path: string, options: RequestInit = {}): Promise<T> {
const res = await fetch(`${this.baseUrl}/api/v1${path}`, {
...options,
headers: {
"Content-Type": "application/json",
Authorization: `token ${this.token}`,
...(options.headers as Record<string, string> | undefined),
},
});
let token = await this.tokens.current();
let res = await this.send(path, options, token);
if (res.status === 401) {
token = await this.tokens.renew(token);
res = await this.send(path, options, token);
}
if (!res.ok) throw new Error(`Gitea API ${path}: ${res.status} ${await res.text()}`);
if (res.status === 204) return null as T;
const text = await res.text();
return (text ? JSON.parse(text) : null) as T;
}
private send(path: string, options: RequestInit, token: string): Promise<Response> {
return fetch(`${this.baseUrl}/api/v1${path}`, {
...options,
headers: {
"Content-Type": "application/json",
Authorization: `token ${token}`,
...(options.headers as Record<string, string> | undefined),
},
});
}
/** Generic authenticated API call — the escape hatch for endpoints without a dedicated method.
* Path is relative to /api/v1. */
async api<T = unknown>(path: string, options: RequestInit = {}): Promise<T> {
@@ -335,24 +352,34 @@ export class GiteaAdmin {
GiteaAdmin.fail("/orgs", res);
}
/** Ensure the org's package team exists, granting read+write on packages, and return its id. The
* team is found by name if it is already there, created otherwise; a lost create race is resolved
* by re-listing. */
/** Ensure the org's package team exists with exactly these units, and return its id. Found or
* created, the units are applied either way — a team is configuration the reconcile loop owns,
* the same as a user's password, so a unit this code gains reaches a team that already exists
* rather than only the next mesh raised from scratch. A lost create race is resolved by
* re-listing. */
async ensureTeam(org: string, team: string, packageWrite: boolean): Promise<number> {
// The units a consumer needs, and no more. `units_map` is exhaustive — a unit not named is a
// unit the team does not have — so code read must be said here: without it gitea answers a
// member's clone of a private repository with "not found", which is how the builder's first
// credentialed clone failed against a team that named only packages.
const units = {
permission: "read",
units_map: { "repo.code": "read", "repo.packages": packageWrite ? "write" : "read" },
includes_all_repositories: true,
can_create_org_repo: false,
};
const found = await this.findTeam(org, team);
if (found !== null) return found;
if (found !== null) {
const patch = await this.request(`/teams/${found}`, {
method: "PATCH",
body: JSON.stringify({ name: team, ...units }),
});
if (patch.status === 200) return found;
GiteaAdmin.fail(`/teams/${found}`, patch);
}
const res = await this.request(`/orgs/${encodeURIComponent(org)}/teams`, {
method: "POST",
body: JSON.stringify({
name: team,
permission: "read",
// Package access is a per-unit grant; the team needs write on the packages unit and nothing
// else. includes_all_repositories keeps the team's repo view whole without widening its
// repo permission beyond read.
units_map: { "repo.packages": packageWrite ? "write" : "read" },
includes_all_repositories: true,
can_create_org_repo: false,
}),
body: JSON.stringify({ name: team, ...units }),
});
if (res.status === 201) return Number(res.body?.id);
if (res.status === 422 || res.status === 409) {
@@ -363,14 +390,18 @@ export class GiteaAdmin {
}
private async findTeam(org: string, team: string): Promise<number | null> {
const res = await this.request(`/orgs/${encodeURIComponent(org)}/teams`);
const res = await this.request(`/orgs/${encodeURIComponent(org)}/teams?limit=50`);
if (res.status !== 200) return null;
const match = (res.body as any[] | null)?.find((t) => t?.name === team);
return match ? Number(match.id) : null;
}
/** Ensure a user exists with exactly this password. Created if absent; if already there, its
* password is patched — so the mesh minting a new secret takes on the next reconcile. */
* password is patched — so the mesh minting a new secret takes on the next reconcile.
*
* The edit path is taken only when the user actually exists. A 422 from the create is also what
* a plain validation failure returns, and reading it as "already there" made the follow-up edit
* 404 — burying the create's own message, which is the one that says what is actually wrong. */
async ensureUser(username: string, password: string, email: string): Promise<void> {
const res = await this.request("/admin/users", {
method: "POST",
@@ -378,13 +409,18 @@ export class GiteaAdmin {
});
if (res.status === 201) return;
if (res.status === 422 || res.status === 409) {
const patch = await this.request(`/admin/users/${encodeURIComponent(username)}`, {
method: "PATCH",
// login_name is required by the admin edit endpoint; for a local user it is the username.
body: JSON.stringify({ login_name: username, password, must_change_password: false }),
});
if (patch.status === 200) return;
GiteaAdmin.fail(`/admin/users/${username}`, patch);
const seen = await this.request(`/users/${encodeURIComponent(username)}`);
if (seen.status === 200) {
const patch = await this.request(`/admin/users/${encodeURIComponent(username)}`, {
method: "PATCH",
// login_name is required by the admin edit endpoint; for a local user it is the username.
// active and prohibit_login: a deactivated or login-prohibited user is refused like a wrong
// password, so the provisioner's check reports it lost; applying again must undo both.
body: JSON.stringify({ login_name: username, password, must_change_password: false, active: true, prohibit_login: false }),
});
if (patch.status === 200) return;
GiteaAdmin.fail(`/admin/users/${username}`, patch);
}
}
GiteaAdmin.fail("/admin/users", res);
}
@@ -399,6 +435,29 @@ export class GiteaAdmin {
GiteaAdmin.fail(`/teams/${teamId}/members/${username}`, res);
}
/**
* Whether a consumer's user logs in with exactly this password and is still a member of the
* package team. Read-only: the password is checked as the consumer presents it, basic auth on the
* API, and membership through the admin API. `false` for a refused login or a missing member; any
* other answer rejects (novox/hq issue 120).
*/
async holdsTeamMember(org: string, team: string, username: string, password: string): Promise<boolean> {
const me = await fetch(`${this.baseUrl}/api/v1/user`, {
headers: { Authorization: "Basic " + Buffer.from(`${username}:${password}`).toString("base64") },
});
if (me.status === 401 || me.status === 403) return false;
if (me.status !== 200) throw new Error(`Gitea GET /user as ${username}: ${me.status}`);
const teams = await this.request(`/orgs/${encodeURIComponent(org)}/teams?limit=50`);
if (teams.status === 404) return false;
if (teams.status !== 200) GiteaAdmin.fail(`/orgs/${org}/teams`, teams);
const found = (teams.body as { id: number; name: string }[]).find((t) => t.name === team);
if (!found) return false;
const member = await this.request(`/teams/${found.id}/members/${encodeURIComponent(username)}`);
if (member.status === 200 || member.status === 204) return true;
if (member.status === 404) return false;
GiteaAdmin.fail(`/teams/${found.id}/members/${username}`, member);
}
/** Delete a user, purging what they own. A 404 means the mesh already withdrew them — success, not
* an error, so a re-run of remove is safe. */
async deleteUser(username: string): Promise<void> {
+17 -2
View File
@@ -16,7 +16,9 @@
import { emit } from "@novox/mesh-sdk/events";
import { GiteaClient } from "./client.js";
// Without a token there is nothing to watch; log and stay quiet rather than crash the runtime.
// Without a way to a token — configured, or mintable with the admin account (token.ts) — there is
// nothing to watch; log and stay quiet rather than crash the runtime. With one, the first poll mints
// or reuses the token, so the runtime's start also shows what it did about it.
let gitea: GiteaClient | null = null;
try {
gitea = GiteaClient.fromEnv();
@@ -49,8 +51,21 @@ async function pollRepos(client: GiteaClient): Promise<void> {
if (gitea) {
const client = gitea;
// A poll that fails says so once, not once a minute: the same reason repeating (the forge not up
// yet, the admin account refused on a restored forge) is one fact, and a recovery is worth a line.
let failing: string | null = null;
const tick = (fn: () => Promise<void>, everyMs: number): void => {
const run = (): void => void fn().catch((err) => console.error(`[gitea] ${err}`));
const run = (): void =>
void fn()
.then(() => {
if (failing !== null) console.log("[gitea] watching again");
failing = null;
})
.catch((err) => {
const why = err instanceof Error ? err.message : String(err);
if (why !== failing) console.error(`[gitea] not watching until this clears — ${why}`);
failing = why;
});
setInterval(run, everyMs);
run();
};
+59 -27
View File
@@ -11,19 +11,27 @@
"name": "gitea"
},
"route": {
"label": "git",
"port": 3000
"web": {
"label": "git",
"port": 3000
},
"internal-api-refused": {
"label": "git",
"path": "/api/internal",
"deny": true,
"priority": 100000
}
}
},
"binds": {
"postgres-database": "/var/lib/gitea/database.json",
"route": "/var/lib/gitea/route.json"
"postgres-database": "${dir:state}/database.json",
"route": "${dir:state}/route.json"
},
"secrets": {
"postgres-database": "/var/lib/gitea/database.secret",
"postgres-database": "${dir:state}/database.secret",
"secret": {
"internal-token": "/var/lib/gitea/internal-token.secret",
"admin": "/var/lib/gitea/admin.secret"
"internal-token": "${dir:state}/internal-token.secret",
"admin": "${dir:state}/admin.secret"
}
},
"capabilities": [
@@ -42,25 +50,39 @@
"why": "the forge, over http"
},
{
"port": 2222,
"port": 22,
"protocol": "tcp",
"from": "mesh",
"why": "git over ssh. Not 22: the machine's own daemon holds that, and a module does not take it"
"why": "git over ssh, gitea's own unmodified sshd. Published on the machine's own side at 222, the mesh's fixed public convention \u2014 not 22, which the machine's own daemon holds and a module does not take"
}
],
"serves": {
"package-registry": {
"npm-package-registry": {
"scheme": "http",
"port": 3000,
"npm-path": "/api/packages/novox/npm/"
},
"git": {
"scheme": "http",
"port": 3000
}
},
"receives": {
"package-registry": "/var/lib/gitea/grants/mesh.json"
"npm-package-registry": "${dir:grants}/npm.json"
},
"grants": {
"package-registry": "/var/lib/gitea/grants"
"npm-package-registry": "${dir:grants}"
},
"claims": [
{
"name": "npm-package-registry",
"scope": "mesh"
},
{
"name": "git",
"scope": "mesh"
}
],
"own-secrets": {
"broker": "/var/lib/mesh/gitea/broker"
},
@@ -71,29 +93,33 @@
"path": "/var/lib/mesh/gitea",
"mode": "0700"
},
{
"id": "runtime-state",
"type": "directory",
"path": "/var/lib/mesh/gitea/state",
"mode": "0700"
},
{
"id": "state",
"type": "directory",
"path": "/var/lib/gitea",
"mode": "0700"
"mode": "0700",
"place": "."
},
{
"id": "grants",
"type": "directory",
"path": "/var/lib/gitea/grants",
"mode": "0700"
},
{
"id": "server-env",
"type": "file",
"path": "/var/lib/gitea/server.env",
"path": "${dir:state}/server.env",
"mode": "0600",
"content": "GITEA__security__INTERNAL_TOKEN=${secret:internal-token}\nGITEA__database__DB_TYPE=postgres\nGITEA__database__HOST=${bound:postgres-database:at}:${bound:postgres-database:port}\nGITEA__database__NAME=${bound:postgres-database:as}\nGITEA__database__USER=${bound:postgres-database:as}\nGITEA__database__PASSWD=${secret:postgres-database}\n"
},
{
"id": "data",
"type": "directory",
"path": "/services/gitea/gitea",
"mode": "0700",
"owner": "1000:1000"
},
@@ -108,14 +134,14 @@
"USER_GID": "1000"
},
"env-file": [
"/var/lib/gitea/server.env"
"${dir:state}/server.env"
],
"ports": [
"3000",
"2222:22"
"222:22"
],
"volumes": [
"/services/gitea/gitea:/data"
"${dir:data}:/data"
],
"secrets-in-environment": "gitea honours GITEA__database__PASSWD__FILE and GITEA__security__INTERNAL_TOKEN__FILE; convertible, awaiting a bed that proves it"
},
@@ -131,11 +157,11 @@
"MESH_GITEA_ADMIN_USER": "mesh-admin"
},
"env-file": [
"/var/lib/gitea/server.env"
"${dir:state}/server.env"
],
"volumes": [
"/services/gitea/gitea:/data",
"/var/lib/gitea/admin.secret:/run/secrets/admin:ro"
"${dir:data}:/data",
"${dir:state}/admin.secret:/run/secrets/admin:ro"
],
"args": [
"/bin/sh",
@@ -160,8 +186,9 @@
"volumes": [
"/var/lib/mesh/gitea/broker:/run/secrets/broker:ro",
"/var/lib/mesh/gitea/config.json:/run/config/config.json:ro",
"/var/lib/gitea/grants:/var/lib/gitea/grants:ro",
"/var/lib/gitea/admin.secret:/run/secrets/admin:ro"
"${dir:grants}:${dir:grants}:ro",
"${dir:state}/admin.secret:/run/secrets/admin:ro",
"/var/lib/mesh/gitea/state:/run/state"
],
"env": {
"MESH_BROKER_FILE": "/run/secrets/broker",
@@ -169,7 +196,8 @@
"MESH_GITEA_CONFIG_FILE": "/run/config/config.json",
"MESH_GITEA_ADMIN_USER": "mesh-admin",
"MESH_GITEA_ADMIN_PASSWORD_FILE": "/run/secrets/admin",
"MESH_RECEIVES": "/var/lib/gitea/grants/mesh.json"
"MESH_GITEA_STATE_DIR": "/run/state",
"MESH_RECEIVES": "${dir:grants}/npm.json"
},
"artifact": "runtime",
"restart-on": [
@@ -179,7 +207,11 @@
],
"provides": [
{
"name": "package-registry",
"name": "npm-package-registry",
"scope": "mesh"
},
{
"name": "git",
"scope": "mesh"
}
],
+5 -1
View File
@@ -4,8 +4,12 @@
"description": "gitea — git hosting. Its API client, tools and events live here (novox/hq ADR 0039).",
"type": "module",
"private": true,
"scripts": {
"build": "tsc client.ts token.ts index.ts provisioner/index.ts tools/index.ts --module NodeNext --moduleResolution NodeNext --target ES2022 --outDir dist",
"test": "npm run build && node --test --experimental-strip-types 'test/*.test.ts'"
},
"dependencies": {
"@novox/mesh-sdk": "^0.1.0"
"@novox/mesh-sdk": "^0.1.1"
},
"devDependencies": {
"@types/node": "^22.0.0",
+27 -7
View File
@@ -1,9 +1,15 @@
// gitea's provisioner — the adapter that makes gitea a provider of the mesh `package-registry`
// interface. The reconcile loop, the contributions file, and reading the mesh's minted password are
// the sdk harness's; this writes only the per-service half: how gitea creates and removes a
// consumer's npm credential (novox/hq ADR 0048/0076).
// gitea's provisioner — the adapter that makes gitea a provider of the mesh
// `npm-package-registry` interface. The reconcile loop, the contributions file, and reading the
// mesh's minted password are the sdk harness's; this writes only the per-service half: how gitea
// creates and removes a consumer's npm credential (novox/hq ADR 0048/0076).
//
// The `package-registry` interface: a consumer authenticates to the npm registry at
// **A package registry seat is one per ecosystem (novox/hq ADR 0109).** gitea holds the npm seat
// (ADR 0110). Adding cargo or PyPI is adding a provision — another `provides` entry, another
// `receives` path and another registration below — not widening this one. `git`, which gitea also
// provides, mints nothing and so registers nothing here: the mesh's own repositories are public,
// and a clone credential is not yet decided (ADR 0111).
//
// The `npm-package-registry` interface: a consumer authenticates to the npm registry at
// `/api/packages/novox/npm/` with basic auth, as `as` with the password the mesh minted, and can
// read and write packages under the `@novox` scope. The registry's npm owner is the gitea org
// `novox`; a consumer is a gitea *user* placed on that org's package team.
@@ -26,7 +32,11 @@ const PACKAGE_TEAM = "packages";
const gitea = GiteaAdmin.fromEnv();
runProvisioner("package-registry", {
// Where this registration's contributions land comes from $MESH_RECEIVES, never a path written
// here: the mesh writes the file where the manifest's `receives` says, and a second copy of that
// path in code would drift from it. One variable carries one path, so a second registration in this
// module needs the mesh to say where each provision's file is — not yet possible, and not faked.
runProvisioner("npm-package-registry", {
async create(p: Provision): Promise<void> {
// The org and its package team are the same for every consumer; ensuring them per-create is
// idempotent and needs no separate bootstrap step.
@@ -34,11 +44,21 @@ runProvisioner("package-registry", {
const teamId = await gitea.ensureTeam(ORG, PACKAGE_TEAM, true);
// The user carries the consumer's login and the mesh's minted password, set every run so a
// rotation takes. Membership of the package team is what grants read+write on packages.
await gitea.ensureUser(p.as, p.password, `${p.as}@localhost`);
//
// The address is gitea's own convention for one that is not real: its email validation
// requires a dotted domain, so `@localhost` was refused at create — the fault that had this
// grant retrying for a day — while `@noreply.localhost` is the shape gitea itself gives
// hidden addresses.
await gitea.ensureUser(p.as, p.password, `${p.as}@noreply.localhost`);
await gitea.addUserToTeam(teamId, p.as);
},
async remove(p: { as: string }): Promise<void> {
await gitea.deleteUser(p.as);
},
// Asked every minute by the harness: whether the backend still holds this consumer exactly as
// the mesh gave it, so a login lost behind the provisioner's back is made again (novox/hq issue 120).
async holds(p: Provision): Promise<boolean> {
return gitea.holdsTeamMember(ORG, PACKAGE_TEAM, p.as, p.password);
},
});
+313
View File
@@ -0,0 +1,313 @@
// What holds the module to its own token (token.ts; hq issue 100, the forge's tools): minted with
// the delivered admin account on the first call and kept at 0600, reused on the next start, minted
// afresh when the forge rejects it or the kept file is gone, and a refused admin account reported in
// plain words rather than crash-looped. A configured token still wins. And the tools register once
// there is a way to a token at all — before one exists.
//
// The forge is a fake: the four routes the module touches, with the same status codes gitea gives.
// Run against the compiled module (npm test builds first), the way the runtime loads it.
import { test, after } from "node:test";
import assert from "node:assert/strict";
import { createServer, type IncomingMessage, type ServerResponse } from "node:http";
import { mkdtemp, readFile, rm, stat, writeFile } from "node:fs/promises";
import { tmpdir } from "node:os";
import { join } from "node:path";
import { collectTools } from "@novox/mesh-sdk/tools";
import { AdminRefused, MintedToken, TOKEN_SCOPES } from "../dist/token.js";
import { GiteaClient } from "../dist/client.js";
import "../dist/tools/index.js";
const ADMIN = "mesh-admin";
const PASSWORD = "the-vault-minted-this";
// ---- A fake forge: what the module sends, and what gitea would answer. ----
interface Forge {
url: string;
mints: number;
lastScopes: string[] | null;
tokens: Map<string, string>;
admins: Map<string, string>;
close(): Promise<void>;
}
function fakeForge(): Promise<Forge> {
const forge = {
mints: 0,
lastScopes: null as string[] | null,
tokens: new Map<string, string>(), // name -> value
scopesOf: new Map<string, string[]>(), // value -> scopes, so a route can enforce them like gitea does
admins: new Map([[ADMIN, PASSWORD]]),
};
// write:X implies read:X — gitea's own rule (models/auth/access_token_scope.go).
const covers = (scopes: string[], required: string): boolean =>
scopes.includes(required) || scopes.includes(`write:${required.split(":")[1]}`);
const json = (res: ServerResponse, status: number, body: unknown): void => {
res.writeHead(status, { "Content-Type": "application/json" });
res.end(body === null ? "" : JSON.stringify(body));
};
const body = (req: IncomingMessage): Promise<any> =>
new Promise((resolve) => {
let text = "";
req.on("data", (c) => (text += c));
req.on("end", () => resolve(text ? JSON.parse(text) : null));
});
const basic = (req: IncomingMessage): string | null => {
const h = req.headers.authorization ?? "";
if (!h.startsWith("Basic ")) return null;
const [user, pass] = Buffer.from(h.slice(6), "base64").toString().split(":");
return forge.admins.get(user) === pass ? user : null;
};
const server = createServer(async (req, res) => {
const url = new URL(req.url ?? "/", "http://fake");
const tokens = url.pathname.match(/^\/api\/v1\/users\/([^/]+)\/tokens(?:\/([^/]+))?$/);
if (tokens) {
const user = basic(req);
if (user === null || user !== decodeURIComponent(tokens[1])) return json(res, 401, { message: "auth required" });
if (req.method === "POST") {
const { name, scopes } = await body(req);
if (forge.tokens.has(name)) return json(res, 400, { message: "token name has already been used" });
forge.mints++;
forge.lastScopes = scopes;
const sha1 = `minted-${forge.mints}-${Math.random().toString(36).slice(2)}`;
forge.tokens.set(name, sha1);
forge.scopesOf.set(sha1, scopes);
return json(res, 201, { id: forge.mints, name, sha1, scopes, token_last_eight: sha1.slice(-8) });
}
if (req.method === "DELETE" && tokens[2]) {
const name = decodeURIComponent(tokens[2]);
if (!forge.tokens.has(name)) return json(res, 404, { message: "token not found" });
forge.tokens.delete(name);
return json(res, 204, null);
}
return json(res, 405, { message: "method not allowed" });
}
if (url.pathname === "/api/v1/user/repos") {
const h = req.headers.authorization ?? "";
const value = h.startsWith("token ") ? h.slice(6) : "";
if (![...forge.tokens.values()].includes(value)) return json(res, 401, { message: "token is required" });
// gitea 1.27.3: GET /user/repos sits under the `user` scope category, not `repository` —
// confirmed against the live forge. A token without read:user (or write:user) is refused here.
const scopes = forge.scopesOf.get(value) ?? [];
if (!covers(scopes, "read:user")) {
return json(res, 403, {
message: `token does not have at least one of required scope(s), required=[read:user]`,
});
}
return json(res, 200, [
{ full_name: "novox/hq", name: "hq", owner: { login: "novox" }, private: true, html_url: "http://fake/novox/hq" },
]);
}
return json(res, 404, { message: "no such route in the fake" });
});
return new Promise((resolve) => {
server.listen(0, "127.0.0.1", () => {
const { port } = server.address() as { port: number };
resolve({
url: `http://127.0.0.1:${port}`,
get mints() { return forge.mints; },
get lastScopes() { return forge.lastScopes; },
tokens: forge.tokens,
admins: forge.admins,
close: () => new Promise((r) => server.close(() => r())),
});
});
});
}
// ---- What the runtime's environment gives the module. ----
async function delivered(forge: Forge): Promise<{ env: NodeJS.ProcessEnv; file: string; logs: string[] }> {
const dir = await mkdtemp(join(tmpdir(), "gitea-"));
const passwordFile = join(dir, "admin.secret");
await writeFile(passwordFile, PASSWORD + "\n", { mode: 0o600 });
const state = join(dir, "state");
return {
env: {
MESH_GITEA_URL: forge.url,
MESH_GITEA_ADMIN_USER: ADMIN,
MESH_GITEA_ADMIN_PASSWORD_FILE: passwordFile,
MESH_GITEA_STATE_DIR: state,
},
file: join(state, "token"),
logs: [],
};
}
/** A client as a fresh process would build it: a new source over the kept file, its log captured. */
function minted(env: NodeJS.ProcessEnv, logs: string[]): GiteaClient {
const source = new MintedToken({
url: env.MESH_GITEA_URL!,
admin: env.MESH_GITEA_ADMIN_USER!,
passwordFile: env.MESH_GITEA_ADMIN_PASSWORD_FILE!,
file: join(env.MESH_GITEA_STATE_DIR!, "token"),
log: (l) => logs.push(l),
});
return new GiteaClient(env.MESH_GITEA_URL!, source);
}
const forge = await fakeForge();
after(() => forge.close());
test("first start: mints with the admin account, keeps the token at 0600, asks for two scopes only", async () => {
const { env, file, logs } = await delivered(forge);
const repos = await minted(env, logs).listRepos();
assert.equal(repos[0]?.full_name, "novox/hq");
assert.equal(forge.mints, 1);
assert.deepEqual(forge.lastScopes, ["write:repository", "write:issue", "read:user"]);
assert.deepEqual(forge.lastScopes, [...TOKEN_SCOPES]);
const token = forge.tokens.get("mesh-tools")!;
assert.equal(await readFile(file, "utf8"), token + "\n");
assert.equal((await stat(file)).mode & 0o777, 0o600);
// Said that it minted, and where it keeps it — never what it is.
assert.ok(logs.some((l) => l.startsWith("minted a token")), logs.join("\n"));
assert.ok(logs.every((l) => !l.includes(token) && !l.includes(PASSWORD)), logs.join("\n"));
});
test("second start: reuses the kept token, mints nothing", async () => {
const { env, logs } = await delivered(forge);
await minted(env, logs).listRepos();
const before = forge.mints;
const again: string[] = [];
await minted(env, again).listRepos();
assert.equal(forge.mints, before);
assert.ok(again.some((l) => l.startsWith("reusing the token kept at")), again.join("\n"));
assert.ok(again.every((l) => !l.includes(forge.tokens.get("mesh-tools")!)), again.join("\n"));
});
test("the forge rejects the kept token (its data was restored): minted afresh, once, and the call goes through", async () => {
const { env, file, logs } = await delivered(forge);
const client = minted(env, logs);
await client.listRepos();
const before = forge.mints;
forge.tokens.clear(); // the forge no longer knows any token — a restore from the predecessor
const repos = await client.listRepos();
assert.equal(repos.length, 1);
assert.equal(forge.mints, before + 1);
assert.equal(await readFile(file, "utf8"), forge.tokens.get("mesh-tools") + "\n");
assert.ok(logs.some((l) => l.startsWith("the forge rejected the kept token")), logs.join("\n"));
});
test("the kept file is gone but the forge still holds a token by that name: replaced, not refused", async () => {
const { env, file, logs } = await delivered(forge);
await minted(env, logs).listRepos();
const before = forge.mints;
await rm(file);
const repos = await minted(env, logs).listRepos();
assert.equal(repos.length, 1);
assert.equal(forge.mints, before + 1);
assert.equal([...forge.tokens.keys()].filter((n) => n === "mesh-tools").length, 1);
assert.ok(logs.some((l) => l.includes('already holds a token named "mesh-tools"')), logs.join("\n"));
});
test("concurrent first calls share one mint", async () => {
const { env, logs } = await delivered(forge);
const client = minted(env, logs);
const before = forge.mints;
await Promise.all([client.listRepos(), client.listRepos(), client.listRepos()]);
assert.equal(forge.mints, before + 1);
});
test("the admin account is refused: said plainly, nothing kept, and the next call fails the same way rather than crashing", async () => {
const { env, file, logs } = await delivered(forge);
forge.admins.delete(ADMIN); // the forge's data came from a predecessor; mesh-admin was never created there
try {
const client = minted(env, logs);
const before = forge.mints;
await assert.rejects(client.listRepos(), (err: unknown) => {
assert.ok(err instanceof AdminRefused, String(err));
assert.match(err.message, /refused the admin account "mesh-admin" \(401\)/);
assert.match(err.message, /admin-bootstrap step creates it/);
assert.match(err.message, /came from a predecessor/);
assert.ok(!err.message.includes(PASSWORD));
return true;
});
await assert.rejects(client.listRepos(), AdminRefused);
assert.equal(forge.mints, before);
await assert.rejects(stat(file), /ENOENT/);
// The account appears (the operator created it): the very next call mints and works.
forge.admins.set(ADMIN, PASSWORD);
assert.equal((await client.listRepos()).length, 1);
assert.equal(forge.mints, before + 1);
} finally {
forge.admins.set(ADMIN, PASSWORD);
}
});
test("one process shares one source per kept file — the watcher and the tools never renew against each other", async () => {
const { env } = await delivered(forge);
assert.equal(MintedToken.fromEnv(env.MESH_GITEA_URL!, env), MintedToken.fromEnv(env.MESH_GITEA_URL!, env));
});
test("a second process finds the token the first renewed, and reuses it instead of minting over it", async () => {
const { env, logs } = await delivered(forge);
const first = minted(env, logs);
const second = minted(env, logs);
await first.listRepos();
await second.listRepos(); // both hold the same kept token
const before = forge.mints;
forge.tokens.clear();
await first.listRepos(); // renews: one mint
await second.listRepos(); // rejected too — but the kept file already carries the renewed one
assert.equal(forge.mints, before + 1);
assert.ok(logs.some((l) => l.includes("is newer — reusing it")), logs.join("\n"));
});
test("a configured token wins, and is reported rather than minted over when the forge rejects it", async () => {
const { env } = await delivered(forge);
const before = forge.mints;
const client = GiteaClient.fromEnv({ ...env, MESH_GITEA_TOKEN: "one-somebody-pasted-in" });
await assert.rejects(client.listRepos(), /rejected the configured Gitea token \(401\)/);
assert.equal(forge.mints, before);
});
test("nothing to mint with and no token: the client says what is missing", async () => {
assert.throws(
() => GiteaClient.fromEnv({ MESH_GITEA_URL: forge.url, MESH_GITEA_ADMIN_USER: ADMIN }),
/set MESH_GITEA_TOKEN, or MESH_GITEA_ADMIN_PASSWORD_FILE, MESH_GITEA_STATE_DIR/,
);
});
test("the tools register once there is a way to a token, and the first call mints it", async () => {
const { env } = await delivered(forge);
const before = forge.mints;
const withAdmin = collectTools(env).find((c) => c.module === "gitea")!;
const withNothing = collectTools({}).find((c) => c.module === "gitea")!;
assert.equal(withNothing.tools.length, 0);
assert.deepEqual(
withAdmin.tools.map((t) => t.name),
[
"gitea_list_repos", "gitea_create_repo", "gitea_delete_repo",
"gitea_list_issues", "gitea_get_issue", "gitea_create_issue", "gitea_close_issue", "gitea_add_comment",
"gitea_list_pull_requests", "gitea_get_pull_request", "gitea_create_pull_request", "gitea_merge_pull_request",
"gitea_list_labels", "gitea_create_label",
"gitea_api",
],
);
assert.equal(forge.mints, before, "registering must not mint — the forge may not be up yet");
const result = (await withAdmin.tools.find((t) => t.name === "gitea_list_repos")!.run({})) as { repos: unknown[] };
assert.equal(result.repos.length, 1);
assert.equal(forge.mints, before + 1);
});
+253
View File
@@ -0,0 +1,253 @@
// The token the forge's tools and watcher authenticate with — and where it comes from.
//
// Nobody configures it. The forge is raised by the mesh, so there is no operator holding a token to
// paste in, and pasting one into settings would put a secret in the inventory in plaintext. What
// the mesh does deliver is the admin account: a login the manifest names and a password the vault
// minted and the host unsealed into a file (novox/hq ADR 0086). That account is enough to mint a
// token, so the module mints its own (hq issue 100, the forge's tools):
//
// - at first use, when none is kept: POST /users/{admin}/tokens over basic auth, with the two
// scopes the tools and the watcher need, and no more;
// - kept in the module's own state, a 0600 file, and read back on the next start — the forge
// hands a token's value out exactly once, so a token not kept is a token lost;
// - re-minted when the forge rejects it (401) or the kept file is gone. The one case that is not
// a fault: the forge's data was restored from a predecessor and the token the file names never
// existed there.
//
// An explicitly configured token still wins, and is never minted over: if it is rejected, that is
// reported, not repaired — somebody chose it.
//
// The token is never logged. Lines say that one was minted, reused or renewed, and where it is
// kept; never what it is.
import { chmodSync, mkdirSync, readFileSync, renameSync, writeFileSync } from "node:fs";
import { dirname, join } from "node:path";
/** The name the token carries in the forge's own list — one per mesh runtime, found by name. */
export const TOKEN_NAME = "mesh-tools";
/**
* The least the fifteen tools and the watcher need (gitea's route groups, 1.20+ scoped tokens):
* write:repository — create/delete repositories, pull requests (list/get/open/merge);
* write:issue — issues, comments, labels;
* read:user — GET /user/repos, which the watcher's poll and gitea_list_repos both call.
* It sits under the `user` category despite listing repositories, not `repository`
* — confirmed against the running forge (1.27.3), which answered
* `required=[read:user]` to a token carrying only the other two.
* Nothing under /admin, /orgs or write:user — the escape-hatch tool reaches only what these three cover.
*/
export const TOKEN_SCOPES: readonly string[] = ["write:repository", "write:issue", "read:user"];
/** Where a client's token comes from, and what to do when the forge says it is wrong. */
export interface TokenSource {
/** The token to authenticate with now; minted, read or configured. */
current(): Promise<string>;
/** The forge answered 401 to `rejected`. A fresh token, or a plain error when there is nothing to renew with. */
renew(rejected: string): Promise<string>;
}
/** A token somebody set — in settings or the environment. Never minted over. */
export class ConfiguredToken implements TokenSource {
constructor(private readonly token: string) {}
async current(): Promise<string> {
return this.token;
}
async renew(): Promise<string> {
throw new Error(
"the forge rejected the configured Gitea token (401). It was set explicitly (settings or MESH_GITEA_TOKEN), " +
"so the module does not mint over it — fix it, or unset it and the module mints its own",
);
}
}
/** The forge would not take the admin account: it is missing, or its password is not the one the mesh holds. */
export class AdminRefused extends Error {
constructor(admin: string, status: number) {
super(
`the forge refused the admin account "${admin}" (${status}) — it does not exist there, or its password is not ` +
`the one the vault delivered. The admin-bootstrap step creates it on a forge the mesh raised; a forge whose data ` +
`came from a predecessor does not have it. Create "${admin}" on the forge with the delivered password and the ` +
`token is minted on the next call — the tools stay registered and the watcher keeps trying`,
);
this.name = "AdminRefused";
}
}
export interface MintedTokenOptions {
/** The forge, e.g. http://127.0.0.1:3000. */
readonly url: string;
/** The admin login the manifest names. */
readonly admin: string;
/** The file the host unsealed the admin password into (ADR 0086). Read at mint time, so a rotation takes. */
readonly passwordFile: string;
/** Where the token is kept: a 0600 file in the module's own state. */
readonly file: string;
readonly name?: string;
readonly scopes?: readonly string[];
readonly log?: (line: string) => void;
readonly fetch?: typeof fetch;
}
/** The token the module mints for itself, kept in its state and renewed when the forge rejects it. */
export class MintedToken implements TokenSource {
private held: string | null = null;
private readFile = false;
private inflight: Promise<string> | null = null;
private readonly name: string;
private readonly scopes: readonly string[];
private readonly log: (line: string) => void;
private readonly fetchImpl: typeof fetch;
constructor(private readonly opts: MintedTokenOptions) {
this.name = opts.name ?? TOKEN_NAME;
this.scopes = opts.scopes ?? TOKEN_SCOPES;
this.log = opts.log ?? ((line) => console.log(`[gitea] ${line}`));
this.fetchImpl = opts.fetch ?? fetch;
}
/**
* Build from the runtime's environment: the forge's URL, the admin login and password file the
* manifest hands the runtime, and the module's state directory (MESH_GITEA_STATE_DIR, a directory
* the runtime mounts writable). Throws, naming what is missing, rather than hand back a source
* that cannot mint.
*/
static fromEnv(url: string, env: NodeJS.ProcessEnv = process.env): MintedToken {
const opts = MintedToken.optionsFromEnv(url, env);
// One source per kept file in a process. The watcher and the tools entrypoint both build a
// client in the same runtime; two sources over one file would each renew on a 401 and drop the
// other's token by name, forever. Shared, a renewal is one renewal.
const shared = MintedToken.shared.get(opts.file);
if (shared) return shared;
const source = new MintedToken(opts);
MintedToken.shared.set(opts.file, source);
return source;
}
private static readonly shared = new Map<string, MintedToken>();
private static optionsFromEnv(url: string, env: NodeJS.ProcessEnv): MintedTokenOptions {
const admin = env.MESH_GITEA_ADMIN_USER;
const passwordFile = env.MESH_GITEA_ADMIN_PASSWORD_FILE;
const stateDir = env.MESH_GITEA_STATE_DIR;
const missing = [
admin ? null : "MESH_GITEA_ADMIN_USER",
passwordFile ? null : "MESH_GITEA_ADMIN_PASSWORD_FILE",
stateDir ? null : "MESH_GITEA_STATE_DIR",
].filter((v): v is string => v !== null);
if (missing.length) {
throw new Error(`no Gitea token, and nothing to mint one with — set MESH_GITEA_TOKEN, or ${missing.join(", ")}`);
}
return { url, admin: admin!, passwordFile: passwordFile!, file: join(stateDir!, "token") };
}
async current(): Promise<string> {
if (this.held !== null) return this.held;
if (!this.readFile) {
this.readFile = true;
const kept = this.read();
if (kept !== null) {
this.held = kept;
this.log(`reusing the token kept at ${this.opts.file}`);
return kept;
}
}
return this.mint("no token kept — minting one");
}
async renew(rejected: string): Promise<string> {
// Another caller already renewed while this one was in flight with the old token.
if (this.held !== null && this.held !== rejected) return this.held;
// Or another process did, and kept it: use what is kept before minting over it.
const kept = this.read();
if (kept !== null && kept !== rejected) {
this.held = kept;
this.log(`the forge rejected the token held; the kept one at ${this.opts.file} is newer — reusing it`);
return kept;
}
this.held = null;
return this.mint("the forge rejected the kept token — minting a fresh one");
}
/** One mint at a time: concurrent first calls share it, rather than each minting its own. */
private mint(why: string): Promise<string> {
if (this.inflight === null) {
this.log(why);
this.inflight = this.doMint().finally(() => {
this.inflight = null;
});
}
return this.inflight;
}
private read(): string | null {
try {
const token = readFileSync(this.opts.file, "utf8").replace(/\n$/, "");
return token.length ? token : null;
} catch (err) {
if ((err as NodeJS.ErrnoException).code === "ENOENT") return null;
throw new Error(`cannot read the kept Gitea token at ${this.opts.file}: ${(err as Error).message}`);
}
}
/** Write the token at 0600, whole or not at all: a temp file beside it, then a rename. */
private keep(token: string): void {
mkdirSync(dirname(this.opts.file), { recursive: true, mode: 0o700 });
const tmp = `${this.opts.file}.tmp`;
writeFileSync(tmp, token + "\n", { mode: 0o600 });
chmodSync(tmp, 0o600);
renameSync(tmp, this.opts.file);
}
private async doMint(): Promise<string> {
let password: string;
try {
password = readFileSync(this.opts.passwordFile, "utf8").replace(/\n$/, "");
} catch (err) {
throw new Error(`cannot read the admin password at ${this.opts.passwordFile}: ${(err as Error).message}`);
}
const authorization = "Basic " + Buffer.from(`${this.opts.admin}:${password}`).toString("base64");
const tokens = `${this.opts.url.replace(/\/+$/, "")}/api/v1/users/${encodeURIComponent(this.opts.admin)}/tokens`;
const call = async (method: string, path = "", body?: unknown): Promise<{ status: number; body: any }> => {
const res = await this.fetchImpl(tokens + path, {
method,
headers: { "Content-Type": "application/json", Authorization: authorization },
...(body === undefined ? {} : { body: JSON.stringify(body) }),
});
const text = await res.text();
let parsed: any = null;
if (text) {
try { parsed = JSON.parse(text); } catch { parsed = text; }
}
return { status: res.status, body: parsed };
};
let res = await call("POST", "", { name: this.name, scopes: this.scopes });
if (res.status === 401 || res.status === 403) throw new AdminRefused(this.opts.admin, res.status);
if (res.status === 400 || res.status === 422) {
// The forge still holds a token by this name whose value we no longer have — the kept file
// went while the forge's data stayed. It is ours to replace: drop it by name and mint again.
this.log(`the forge already holds a token named "${this.name}" — replacing it`);
const dropped = await call("DELETE", `/${encodeURIComponent(this.name)}`);
if (dropped.status !== 204 && dropped.status !== 404) {
throw new Error(`Gitea DELETE /users/${this.opts.admin}/tokens/${this.name}: ${dropped.status} ${detail(dropped.body)}`);
}
res = await call("POST", "", { name: this.name, scopes: this.scopes });
}
if (res.status !== 201 && res.status !== 200) {
throw new Error(`Gitea POST /users/${this.opts.admin}/tokens: ${res.status} ${detail(res.body)}`);
}
const token = typeof res.body?.sha1 === "string" ? res.body.sha1 : null;
if (!token) throw new Error(`Gitea POST /users/${this.opts.admin}/tokens: ${res.status} but no token in the reply`);
this.keep(token);
this.held = token;
this.log(`minted a token for "${this.opts.admin}" (${this.scopes.join(", ")}), kept at ${this.opts.file}`);
return token;
}
}
function detail(body: unknown): string {
return typeof body === "string" ? body : JSON.stringify(body);
}
+6 -3
View File
@@ -295,12 +295,15 @@ export function getGiteaTools(gitea: GiteaClient): ToolDefinition[] {
];
}
// The tools exist only when a token can be found; without one, gitea contributes none rather than
// failing the whole runtime.
// The tools exist when the client has a way to a token: one configured, or the admin account to mint
// one with (token.ts). The mint itself happens on the first call, not here — a contributor is
// synchronous, and a forge not yet answering must not keep the runtime from serving. Without either
// way, gitea contributes none rather than failing the whole runtime, and says why.
registerModuleTools("gitea", (env) => {
try {
return getGiteaTools(GiteaClient.fromEnv(env));
} catch {
} catch (err) {
console.log(`[gitea] no tools — ${err instanceof Error ? err.message : String(err)}`);
return [];
}
});
+1 -1
View File
@@ -8,5 +8,5 @@
"skipLibCheck": true,
"noEmit": true
},
"include": ["client.ts", "index.ts", "provisioner/index.ts", "tools/index.ts"]
"include": ["client.ts", "token.ts", "index.ts", "provisioner/index.ts", "tools/index.ts"]
}
+1 -1
View File
@@ -50,7 +50,7 @@
"name": "hello-web",
"network": "hello-web",
"ports": [
"8080:8080"
"8080"
],
"volumes": [
"/var/lib/hello-web/index.html:/www/index.html:ro"
+9 -6
View File
@@ -14,12 +14,15 @@
"mongodb-database": {
"name": "invoicing"
},
"s3-bucket": {
"bucket": "invoicing"
},
"route": {
"label": "invoicing",
"port": 80
"site": {
"label": "invoicing",
"port": 80
},
"api": {
"label": "invoicing-api",
"port": 9000
}
}
},
"binds": {
@@ -63,7 +66,7 @@
"type": "file",
"path": "/var/lib/invoicing/api.env",
"mode": "0600",
"content": "NODE_ENV=production\nPORT=9000\nMONGO_URL=mongodb://${bound:mongodb-database:as}:${secret:mongodb-database}@${bound:mongodb-database:at}:${bound:mongodb-database:port}/invoicing?authSource=admin\nMINIO_BUCKET=invoicing\nMINIO_ENDPOINT=${bound:s3-bucket:at}\nMINIO_PORT=${bound:s3-bucket:port}\nMINIO_ACCESSKEY=${bound:s3-bucket:as}\nMINIO_SECRET=${secret:s3-bucket}\n"
"content": "NODE_ENV=production\nPORT=9000\nMONGO_URL=mongodb://${bound:mongodb-database:as}:${secret:mongodb-database}@${bound:mongodb-database:at}:${bound:mongodb-database:port}/${bound:mongodb-database:as}?authSource=${bound:mongodb-database:as}\nMONGO_DB=${bound:mongodb-database:as}\nMINIO_BUCKET=mesh-novox-invoice\nMINIO_ENDPOINT=${bound:s3-bucket:at}\nMINIO_PORT=${bound:s3-bucket:port}\nMINIO_ACCESSKEY=${bound:s3-bucket:as}\nMINIO_SECRET=${secret:s3-bucket}\n"
},
{
"id": "net",
+4 -2
View File
@@ -88,7 +88,9 @@
"env": {
"KC_DB": "postgres",
"KC_HTTP_ENABLED": "true",
"KC_HEALTH_ENABLED": "true"
"KC_HEALTH_ENABLED": "true",
"KC_HOSTNAME": "https://keycloak.novox.be",
"KC_PROXY_HEADERS": "xforwarded"
},
"env-file": [
"/var/lib/keycloak/admin.env",
@@ -118,7 +120,7 @@
],
"env": {
"MESH_BROKER_FILE": "/run/secrets/broker",
"MESH_KEYCLOAK_URL": "http://127.0.0.1:8080",
"MESH_KEYCLOAK_URL": "http://127.0.0.1:${port:8080}",
"MESH_KEYCLOAK_CONFIG_FILE": "/run/config/config.json"
},
"restart-on": [
+33
View File
@@ -94,6 +94,39 @@ export class LavinmqClient {
await this.api("PUT", `/permissions/${v}/${u}`, { configure: ".*", write: ".*", read: ".*" });
}
/**
* Whether a consumer's user exists with exactly this password and full permissions on its own
* vhost. Read-only: the stored hash is salted SHA-256, the scheme `rabbitHash` writes, so the
* password is checked by hashing it with the stored salt rather than by logging in. `false` when
* the user or its permission is gone or the password differs; an unreachable API rejects
* (novox/hq issue 120).
*/
async holdsConsumer(login: string, password: string): Promise<boolean> {
const v = encodeURIComponent(login);
const u = encodeURIComponent(login);
const user = await this.getOrNull<{ password_hash?: string; hashing_algorithm?: string }>(`/users/${u}`);
if (!user?.password_hash) return false;
if (user.hashing_algorithm && !/sha256/i.test(user.hashing_algorithm)) {
throw new Error(`lavinmq user ${login} is hashed with ${user.hashing_algorithm}, which this check cannot verify`);
}
const stored = Buffer.from(user.password_hash, "base64");
if (stored.length < 5 || rabbitHash(password, stored.subarray(0, 4)) !== user.password_hash) return false;
const perm = await this.getOrNull<{ configure?: string; write?: string; read?: string }>(`/permissions/${v}/${u}`);
return perm?.configure === ".*" && perm?.write === ".*" && perm?.read === ".*";
}
/** A GET that answers null for a 404 and rejects on anything else that is not 2xx. */
private async getOrNull<T>(path: string): Promise<T | null> {
const resp = await fetch(`${this.conn.base}/api${path}`, {
headers: {
Authorization: "Basic " + Buffer.from(`${this.conn.adminUser}:${this.conn.adminPassword}`).toString("base64"),
},
});
if (resp.status === 404) return null;
if (!resp.ok) throw new Error(`lavinmq management API GET ${path} -> ${resp.status}: ${await resp.text()}`);
return (await resp.json()) as T;
}
/** Remove a consumer's vhost and user, idempotently. A DELETE of what is already gone is tolerated. */
async removeConsumer(login: string): Promise<void> {
const v = encodeURIComponent(login);
+15 -3
View File
@@ -75,6 +75,12 @@
"path": "/var/lib/lavinmq-module/grants",
"mode": "0700"
},
{
"id": "broker-data",
"type": "directory",
"path": "/var/lib/mesh-broker",
"mode": "0700"
},
{
"id": "server",
"type": "container",
@@ -86,8 +92,8 @@
"127.0.0.1:15672:15672"
],
"volumes": [
"mesh-broker-data:/var/lib/lavinmq",
"mesh-broker-tls:/tls:ro"
"/var/lib/mesh-broker:/var/lib/lavinmq",
"/var/lib/mesh-broker-tls:/tls:ro"
],
"args": [
"--amqps-port=5671",
@@ -135,5 +141,11 @@
"from": "Dockerfile"
}
]
}
},
"accesses": [
{
"path": "/var/lib/mesh-broker-tls",
"mode": "read"
}
]
}
+1 -1
View File
@@ -5,7 +5,7 @@
"type": "module",
"private": true,
"dependencies": {
"@novox/mesh-sdk": "^0.1.0"
"@novox/mesh-sdk": "^0.1.1"
},
"devDependencies": {
"@types/node": "^22.0.0",
+5
View File
@@ -48,4 +48,9 @@ runProvisioner("amqp", {
await lavinmq.removeConsumer(p.as);
await announce("module.lavinmq.amqp.deprovisioned", { user: p.as, vhost: p.as });
},
// Asked every minute by the harness: whether the backend still holds this consumer exactly as
// the mesh gave it, so a login lost behind the provisioner's back is made again (novox/hq issue 120).
async holds(p: Provision): Promise<boolean> {
return lavinmq.holdsConsumer(p.as, p.password);
},
});
+2 -2
View File
@@ -17,7 +17,7 @@ FROM ${BUILD_BASE} AS build
# resolved away.
WORKDIR /app/modules/mailu
COPY . .
RUN node /app/node_modules/typescript/bin/tsc client.ts index.ts tools/index.ts \
RUN node /app/node_modules/typescript/bin/tsc client.ts index.ts tools/index.ts provisioner/index.ts \
--module NodeNext --moduleResolution NodeNext --target ES2022 --outDir dist
FROM ${RUNTIME_BASE}
@@ -27,4 +27,4 @@ COPY --from=build /app/modules/mailu/dist /app/modules/mailu/dist
# the convention novox/hq issues 060/061 settled. A container that instead ran only its
# provisioner (`run`) served no tools and emitted no events; a container that named no command
# ran no provisioner at all.
ENV MESH_TOOL_MODULES=/app/modules/mailu/dist/index.js,/app/modules/mailu/dist/tools/index.js
ENV MESH_TOOL_MODULES=/app/modules/mailu/dist/index.js,/app/modules/mailu/dist/tools/index.js,/app/modules/mailu/dist/provisioner/index.js
+44
View File
@@ -0,0 +1,44 @@
# automx2 — the autoconfig/autodiscover responder, carried by the mailu module as its own
# artifact: it is a config-baked sidecar of this mail server, not a standalone application
# (novox/hq ADR 0015 draws that line at applications).
#
# The base is named rather than pinned (novox/hq issue 044): declared in module.json's
# `build.on`. The build context is the module's own directory; every ADD says so.
ARG PYTHON_BASE
FROM ${PYTHON_BASE}
RUN apk add --no-cache bash sqlite
WORKDIR /automx2
ADD automx/files/setupvenv.sh /automx2/setupvenv.sh
ADD automx/files/start /automx2/start
ADD automx/files/setup /automx2/setup
ADD automx/files/setup-db /automx2/setup-db
ADD automx/files/add-domains /automx2/add-domains
RUN chmod u+x setupvenv.sh start add-domains setup setup-db
RUN ./setupvenv.sh \
&& . .venv/bin/activate \
&& pip install automx2==2021.6
# The launcher `start` expects. In the predecessor's image this wrapper appeared during a build
# step that never made it into the files this module carries — the image worked and the recipe
# could not reproduce it. Written here explicitly, verbatim from the proven image, so the build
# is the whole truth about the image again.
RUN mkdir -p .venv/scripts && printf '%s\n' \
'#!/usr/bin/env bash' \
'set -euo pipefail' \
'. .venv/bin/activate' \
"export FLASK_ENV='production'" \
"export FLASK_APP='automx2.server:app'" \
'flask "$@"' > .venv/scripts/flask.sh && chmod +x .venv/scripts/flask.sh
ENV AUTOMX2_CONF=/etc/automx2.conf
ADD automx/files/automx2.conf /etc/automx2.conf
# VOLUME deliberately absent: the anonymous /data volume is exactly what lost db.sqlite on
# every recreate (measured on novox 2026-08-10). The manifest binds a real directory instead.
ENTRYPOINT ["/bin/sh"]
CMD ["./start"]
EXPOSE 4243
+49
View File
@@ -0,0 +1,49 @@
#!/usr/bin/env bash
set -e
echo "${MAIL_DOMAINS}"
# Split domains into array
IFS=', ' read -r -a array <<< "${AMX_MAIL_DOMAINS}"
# User configurable section -- START
PROVIDER_ID=001
SQL_CMD="";
# Iterate domains resulting from split on second arg
for element in "${array[@]}"
do
# Set vars
DOMAIN=$element
PROVIDER_NAME=$DOMAIN
PROVIDER_SHORTNAME=$DOMAIN
# Optional LDAP server
#LDAP_SERVER="ldap.${DOMAIN}"
# User configurable section -- END
s1_id=$((PROVIDER_ID + 1))
s2_id=$((PROVIDER_ID + 2))
s3_id=$((PROVIDER_ID + 3))
dom_id=$((PROVIDER_ID + 4))
s3_id='NULL'
SQL_CMD=$(cat <<EOT
$SQL_CMD
INSERT INTO provider(id, name, short_name) VALUES(${PROVIDER_ID}, '${PROVIDER_NAME}', '${PROVIDER_SHORTNAME}');
INSERT INTO server(id, port, type, name, socket_type, user_name, authentication)
VALUES(${s1_id}, ${AMX_IMAP_PORT}, 'imap', '${AMX_IMAP_SERVER}', 'STARTTLS', '%EMAILADDRESS%', 'password-cleartext');
INSERT INTO server(id, port, type, name, socket_type, user_name, authentication)
VALUES(${s2_id}, ${AMX_SMTP_PORT}, 'smtp', '${AMX_SMTP_ADDRESS}', 'STARTTLS', '%EMAILADDRESS%', 'password-cleartext');
INSERT INTO domain(id, name, provider_id, ldapserver_id) VALUES(${dom_id}, '${DOMAIN}', ${PROVIDER_ID}, ${s3_id});
INSERT INTO server_domain(server_id, domain_id) VALUES(${s1_id}, ${dom_id});
INSERT INTO server_domain(server_id, domain_id) VALUES(${s2_id}, ${dom_id});
EOT
)
PROVIDER_ID=$((PROVIDER_ID+10))
done
echo -e ${SQL_CMD}
echo -e ${SQL_CMD} | sqlite3 /data/db.sqlite
+21
View File
@@ -0,0 +1,21 @@
[automx2]
# A typical production setup would use loglevel = WARNING
loglevel = WARNING
# Echo SQL commands into log? Used for debugging.
db_echo = false
# In-memory SQLite database
# db_uri = sqlite:///:memory:
# SQLite database in a UNIX-like file system
db_uri = sqlite:////data/db.sqlite
# MySQL database on a remote server. This example does not use an encrypted
# connection and is therefore *not* recommended for production use.
#db_uri = mysql://username:password@server.example.com/db
# Number of proxy servers between automx2 and the client (default: 0).
# If your logs only show 127.0.0.1 or ::1 as the source IP for incoming
# connections, proxy_count probably needs to be changed.
proxy_count = 1
+12
View File
@@ -0,0 +1,12 @@
#!/usr/bin/env bash
set -e
if [ ! -e /data/db.sqlite ]; then
# DB SETUP
echo "SETTING UP DB"
./setup-db
echo "ADDING DOMAINS"
# Add the domains
./add-domains
fi
+84
View File
@@ -0,0 +1,84 @@
#!/usr/bin/env bash
set -e
# LDAP-Server
LDAP=$(cat <<EOT
CREATE TABLE ldapserver(
id INT PRIMARY KEY NOT NULL,
name TEXT NOT NULL,
port INT NOT NULL,
use_ssl INT NOT NULL,
search_base TEXT NOT NULL,
search_filter TEXT NOT NULL,
attr_uid TEXT NOT NULL,
attr_cn TEXT NOT NULL,
bind_password TEXT NOT NULL,
bind_user TEXT NOT NULL
);
EOT
)
# Provider
PROVIDER=$(cat <<EOT
CREATE TABLE provider(
id INT PRIMARY KEY NOT NULL,
name TEXT NOT NULL,
short_name TEXT NOT NULL
);
EOT
)
# Server
SERVER=$(cat <<EOT
CREATE TABLE server(
id INT PRIMARY KEY NOT NULL,
prio INT NOT NULL DEFAULT 10,
name TEXT NOT NULL,
port INT NOT NULL,
type TEXT NOT NULL,
socket_type TEXT NOT NULL,
user_name TEXT NOT NULL,
authentication TEXT NOT NULL
);
EOT
)
# Domain
DOMAIN=$(cat <<EOT
CREATE TABLE domain(
id INT PRIMARY KEY NOT NULL,
name TEXT NOT NULL,
provider_id INT NOT NULL,
ldapserver_id INT NULL,
FOREIGN KEY(ldapserver_id) REFERENCES ldapserver(id),
FOREIGN KEY(provider_id) REFERENCES provider(id)
);
CREATE UNIQUE INDEX domain_name ON domain(name);
EOT
)
# Server-Domain
SERVER_DOMAIN=$(cat <<EOT
CREATE TABLE server_domain(
server_id INT NOT NULL,
domain_id INT NOT NULL,
FOREIGN KEY(server_id) REFERENCES server(id),
FOREIGN KEY(domain_id) REFERENCES domain(id)
);
EOT
)
## TODO Foreign keys
SQL_CMD=$(cat <<EOT
$LDAP
$PROVIDER
$SERVER
$DOMAIN
$SERVER_DOMAIN
EOT
)
echo -e ${SQL_CMD}
echo -e ${SQL_CMD} | sqlite3 /data/db.sqlite
+38
View File
@@ -0,0 +1,38 @@
#!/usr/bin/env bash
# vim:ts=4:sw=4:noet
#
# Creates a Python 3 virtual environment. The target directory can be passed
# as a parameter. The default path is '.venv' in the current directory.
dir="${1:-.venv}"
echo "Setup dir $dir"
set -e
if [ -d "${dir}" ]; then
echo >&2 "Directory '${dir}' already exists, exiting."
exit 1
fi
python3 -m venv "${dir}"
source "${dir}/bin/activate"
set +e
pip install -U pip setuptools wheel || true
#set -e
## vim:tabstop=4:noexpandtab
##
## Creates a Python 3 virtual environment. The target directory can be passed
## as a parameter. The default path is 'venv' in the current directory.
#
#dir="${1:-venv}"
#
#set -e
#if [ -d "${dir}" ]; then
# echo "Directory '${dir}' already exists, exiting." >&2
# exit 1
#fi
#python3 -m venv "${dir}"
#. "${dir}/bin/activate"
#
#set +e
#pip install -U pip setuptools || true
+8
View File
@@ -0,0 +1,8 @@
#!/usr/bin/env bash
set -e
# Setup
./setup
# Start
./.venv/scripts/flask.sh run --host=0.0.0.0 --port=4243
+29
View File
@@ -130,10 +130,39 @@ export class MailuClient {
await this.api("PATCH", `/user/${encodeURIComponent(email)}`, { raw_password: password });
}
/**
* Set the mesh's password on a mailbox the mesh provisions, and enable it. A disabled mailbox is
* what the provisioner's check reports as lost, so applying again must enable it, or the two would
* disagree for ever. Separate from changePassword, which an operator's tool uses and which must
* not re-enable a mailbox someone disabled.
*/
async applyProvisioned(email: string, password: string): Promise<void> {
await this.api("PATCH", `/user/${encodeURIComponent(email)}`, { raw_password: password, enabled: true });
}
async deleteUser(email: string): Promise<void> {
await this.api("DELETE", `/user/${encodeURIComponent(email)}`);
}
/**
* Whether a mailbox exists and is enabled. Read-only, through the admin API.
*
* **The password is not checked.** Mailu authenticates in its admin service, behind the front;
* the imap server's own password database accepts any password from Mailu's subnet, so asking it
* (`doveadm auth test`) proves nothing, or refuses everyone. A lost or disabled mailbox is caught;
* a password changed by hand is not (novox/hq issue 120).
*/
async holdsUser(email: string): Promise<boolean> {
const res = await fetch(`${this.baseUrl}/user/${encodeURIComponent(email)}`, {
headers: { Authorization: this.apiKey, Accept: "application/json" },
});
if (res.status === 404) return false;
if (!res.ok) throw new Error(`Mailu API GET /user/${email}: ${res.status} ${await res.text()}`);
const user = (await res.json()) as { enabled?: boolean };
return user.enabled !== false;
}
async listAliases(): Promise<MailuAlias[]> {
const aliases = await this.api<any[]>("GET", "/alias");
return (aliases ?? []).map((a) => ({
+218 -89
View File
@@ -14,20 +14,42 @@
"name": "mailu"
},
"route": {
"label": "mail",
"port": 7080
"web": {
"label": "mail",
"port": 7443,
"scheme": "https",
"insecure": true
},
"acme": {
"label": "mail",
"path": "/.well-known/acme-challenge",
"port": 7080,
"priority": 100
},
"autoconfig": {
"label": "autoconfig",
"port": 4243
},
"autodiscover": {
"label": "autodiscover",
"port": 4243
},
"automx": {
"label": "automx",
"port": 4243
}
}
},
"binds": {
"postgres-database": "/var/lib/mailu/database.json",
"route": "/var/lib/mailu/route.json"
"postgres-database": "${dir:state}/database.json",
"route": "${dir:state}/route.json"
},
"secrets": {
"postgres-database": "/var/lib/mailu/database.secret",
"postgres-database": "${dir:state}/database.secret",
"secret": {
"secret-key": "/var/lib/mailu/secret-key.secret",
"admin": "/var/lib/mailu/admin.secret",
"api-token": "/var/lib/mailu/api-token.secret"
"secret-key": "${dir:state}/secret-key.secret",
"admin": "${dir:state}/admin.secret",
"api-token": "${dir:state}/api-token.secret"
}
},
"emits": [
@@ -44,6 +66,20 @@
"why": "mail from other mail servers",
"fixed": true
},
{
"port": 110,
"protocol": "tcp",
"from": "anywhere",
"why": "POP3, kept at parity with the predecessor; pruning legacy protocols is its own deliberate change",
"fixed": true
},
{
"port": 143,
"protocol": "tcp",
"from": "anywhere",
"why": "IMAP with STARTTLS, kept at parity",
"fixed": true
},
{
"port": 465,
"protocol": "tcp",
@@ -55,7 +91,7 @@
"port": 587,
"protocol": "tcp",
"from": "anywhere",
"why": "submission",
"why": "submission; also what the smtp provision serves consumers",
"fixed": true
},
{
@@ -65,11 +101,30 @@
"why": "IMAP over TLS",
"fixed": true
},
{
"port": 995,
"protocol": "tcp",
"from": "anywhere",
"why": "POP3 over TLS, kept at parity",
"fixed": true
},
{
"port": 7080,
"protocol": "tcp",
"from": "mesh",
"why": "the web interface (admin, webmail, admin API), behind the route proxy"
"why": "the web front over http; only the ACME HTTP-01 passthrough is routed here \u2014 everything else 301s to https and would loop a proxy"
},
{
"port": 7443,
"protocol": "tcp",
"from": "mesh",
"why": "the web front over its own TLS (admin, webmail, API); the public name mail.novox.be is a route grant reaching it here"
},
{
"port": 4243,
"protocol": "tcp",
"from": "mesh",
"why": "automx: mail client autoconfiguration; autoconfig/autodiscover/automx.novox.be are route grants reaching it here"
}
],
"own-secrets": {
@@ -85,125 +140,125 @@
{
"id": "state",
"type": "directory",
"path": "/var/lib/mailu",
"mode": "0700",
"place": "."
},
{
"id": "grants",
"type": "directory",
"mode": "0700"
},
{
"id": "data-automx",
"type": "directory",
"mode": "0700"
},
{
"id": "config-env",
"type": "file",
"path": "/var/lib/mailu/mailu.env",
"path": "${dir:state}/mailu.env",
"mode": "0644",
"content": "DOMAIN=novox.be\nHOSTNAMES=mail.novox.be\nPOSTMASTER=admin\nSITENAME=Novox\nWEBSITE=https://novox.be\nTLS_FLAVOR=cert\nSUBNET=192.168.203.0/24\nCOMPOSE_PROJECT_NAME=mailu\nHOST_ADMIN=mailu-admin\nHOST_ANTISPAM=mailu-antispam:11332\nHOST_IMAP=mailu-imap\nHOST_SMTP=mailu-smtp\nHOST_WEBMAIL=mailu-webmail\nHOST_WEBDAV=mailu-webdav:5232\nHOST_REDIS=mailu-redis\nHOST_FRONT=mailu-front\nREDIS_ADDRESS=mailu-redis\nANTIVIRUS=clamav\nWEBMAIL=roundcube\nWEBDAV=radicale\nFETCHMAIL_ENABLED=True\nFETCHMAIL_DELAY=600\nADMIN=true\nWEB_ADMIN=/admin\nWEB_WEBMAIL=/webmail\nWEBROOT_REDIRECT=/webmail\nWEBMAIL_ADDRESS=webmail\nAPI=true\nWEB_API=/api\nAUTH_RATELIMIT_IP=6000/hour\nAUTH_RATELIMIT_USER=1000/day\nCREDENTIAL_ROUNDS=12\nPASSWORD_SCHEME=PBKDF2\nDISABLE_STATISTICS=True\nMESSAGE_SIZE_LIMIT=50000000\nMESSAGE_RATELIMIT=200/day\nRECIPIENT_DELIMITER=+\nPOSTFIX_MYNETWORKS=127.0.0.0/8 [::1]/128\nRELAYNETS=\nRELAYHOST=\nREJECT_UNLISTED_RECIPIENT=\nDB_FLAVOR=postgresql\nINITIAL_ADMIN_ACCOUNT=admin\nINITIAL_ADMIN_DOMAIN=novox.be\nINITIAL_ADMIN_MODE=ifmissing\nSMTP_PORT=25\nSMTPS_PORT=465\nSUBMISSION_PORT=587\nPOP3_PORT=110\nPOP3S_PORT=995\nIMAP_PORT=143\nIMAPS_PORT=993\nHTTP_PORT=7080\nHTTPS_PORT=7443\nAUTOMX_PORT=4243\nAMX_SMTP_ADDRESS=mail.novox.be\nAMX_SMTP_PORT=587\nAMX_IMAP_ADDRESS=mail.novox.be\nAMX_IMAP_PORT=143\nAMX_MAIL_DOMAINS=novox.be\nDMARC_RUA=admin\nDMARC_RUF=admin\nLETSENCRYPT_SHORTCHAIN=True\nTZ=Etc/UTC\nLOG_LEVEL=INFO\nWELCOME=false\nREAL_IP_HEADER=X-Real-IP\nREAL_IP_FROM=\nCOMPRESSION=\nCOMPRESS_LEVEL=\nCOMPRESSION_LEVEL=\nBIND_ADDRESS4=127.0.0.1\nBIND_ADDRESS6=::1\nMAILU_VERSION=1.9\nDOCKER_ORG=mailu\nDOCKER_PREFIX=\n"
"content": "ADMIN_ADDRESS=mailu-admin\nANTISPAM_ADDRESS=mailu-antispam\nANTIVIRUS_ADDRESS=mailu-antivirus\nIMAP_ADDRESS=mailu-imap\nSMTP_ADDRESS=mailu-smtp\nFRONT_ADDRESS=mailu-front\nWEBMAIL_ADDRESS=mailu-webmail\nWEBDAV_ADDRESS=mailu-webdav\nREDIS_ADDRESS=mailu-redis\nPORTS=25,80,443,465,993,995,4190,110,143,587\nDOMAIN=novox.be\nHOSTNAMES=mail.novox.be\nPOSTMASTER=admin\nSITENAME=Novox\nWEBSITE=https://novox.be\nTLS_FLAVOR=letsencrypt\nSUBNET=192.168.203.0/24\nCOMPOSE_PROJECT_NAME=mailu\nANTIVIRUS=clamav\nWEBMAIL=roundcube\nWEBDAV=radicale\nFETCHMAIL_ENABLED=True\nFETCHMAIL_DELAY=600\nADMIN=true\nWEB_ADMIN=/admin\nWEB_WEBMAIL=/webmail\nWEBROOT_REDIRECT=/webmail\nAPI=true\nWEB_API=/api\nAUTH_RATELIMIT_IP=6000/hour\nAUTH_RATELIMIT_USER=1000/day\nCREDENTIAL_ROUNDS=12\nPASSWORD_SCHEME=PBKDF2\nDISABLE_STATISTICS=True\nMESSAGE_SIZE_LIMIT=50000000\nMESSAGE_RATELIMIT=200/day\nRECIPIENT_DELIMITER=+\nPOSTFIX_MYNETWORKS=127.0.0.0/8 [::1]/128\nRELAYNETS=\nRELAYHOST=\nREJECT_UNLISTED_RECIPIENT=\nDB_FLAVOR=postgresql\nINITIAL_ADMIN_ACCOUNT=admin\nINITIAL_ADMIN_DOMAIN=novox.be\nINITIAL_ADMIN_MODE=ifmissing\nSMTP_PORT=25\nSMTPS_PORT=465\nSUBMISSION_PORT=587\nPOP3_PORT=110\nPOP3S_PORT=995\nIMAP_PORT=143\nIMAPS_PORT=993\nHTTP_PORT=7080\nHTTPS_PORT=7443\nAUTOMX_PORT=4243\nAMX_SMTP_ADDRESS=mail.novox.be\nAMX_SMTP_PORT=587\nAMX_IMAP_ADDRESS=mail.novox.be\nAMX_IMAP_PORT=143\nAMX_MAIL_DOMAINS=novox.be\nDMARC_RUA=admin\nDMARC_RUF=admin\nLETSENCRYPT_SHORTCHAIN=True\nTZ=Etc/UTC\nLOG_LEVEL=INFO\nWELCOME=false\nREAL_IP_HEADER=X-Real-IP\nREAL_IP_FROM=142.132.152.141\nCOMPRESSION=\nCOMPRESS_LEVEL=\nCOMPRESSION_LEVEL=\nBIND_ADDRESS4=127.0.0.1\nBIND_ADDRESS6=::1\nMAILU_VERSION=1.9\nDOCKER_ORG=mailu\nDOCKER_PREFIX=\nWELCOME_SUBJECT=Welcome to your new email account\nWELCOME_BODY=Welcome to your new email account, if you can read this, then it is configured properly!\n"
},
{
"id": "secret-env",
"type": "file",
"path": "/var/lib/mailu/secret.env",
"path": "${dir:state}/secret.env",
"mode": "0600",
"content": "SECRET_KEY=${secret:secret-key}\n"
},
{
"id": "database-env",
"type": "file",
"path": "/var/lib/mailu/database.env",
"path": "${dir:state}/database.env",
"mode": "0600",
"content": "DB_FLAVOR=postgresql\nDB_HOST=${bound:postgres-database:at}:${bound:postgres-database:port}\nDB_USER=${bound:postgres-database:as}\nDB_NAME=${bound:postgres-database:as}\nDB_PW=${secret:postgres-database}\n"
},
{
"id": "admin-env",
"type": "file",
"path": "/var/lib/mailu/admin.env",
"path": "${dir:state}/admin.env",
"mode": "0600",
"content": "INITIAL_ADMIN_PW=${secret:admin}\nAPI_TOKEN=${secret:api-token}\n"
},
{
"id": "data-certs",
"type": "directory",
"path": "/services/mailu/data/certs",
"mode": "0700"
},
{
"id": "data-data",
"type": "directory",
"path": "/services/mailu/data/data",
"mode": "0700"
},
{
"id": "data-dkim",
"type": "directory",
"path": "/services/mailu/data/dkim",
"mode": "0700"
},
{
"id": "data-mail",
"type": "directory",
"path": "/services/mailu/data/mail",
"mode": "0700"
},
{
"id": "data-mailqueue",
"type": "directory",
"path": "/services/mailu/data/mailqueue",
"mode": "0700"
"mode": "0755"
},
{
"id": "data-filter",
"type": "directory",
"path": "/services/mailu/data/filter",
"mode": "0700"
},
{
"id": "data-clamav",
"type": "directory",
"mode": "0700"
},
{
"id": "data-redis",
"type": "directory",
"path": "/services/mailu/data/redis",
"mode": "0700"
},
{
"id": "data-webmail",
"type": "directory",
"path": "/services/mailu/data/webmail",
"mode": "0700"
},
{
"id": "data-dav",
"type": "directory",
"path": "/services/mailu/data/dav",
"mode": "0700"
},
{
"id": "data-fetchmail",
"type": "directory",
"path": "/services/mailu/data/data/fetchmail",
"mode": "0700"
},
{
"id": "data-overrides-nginx",
"type": "directory",
"path": "/services/mailu/data/overrides/nginx",
"mode": "0700"
},
{
"id": "data-overrides-dovecot",
"type": "directory",
"path": "/services/mailu/data/overrides/dovecot",
"mode": "0700"
},
{
"id": "data-overrides-postfix",
"type": "directory",
"path": "/services/mailu/data/overrides/postfix",
"mode": "0700"
},
{
"id": "data-overrides-rspamd",
"type": "directory",
"path": "/services/mailu/data/overrides/rspamd",
"mode": "0700"
},
{
"id": "data-overrides-roundcube",
"type": "directory",
"path": "/services/mailu/data/overrides/roundcube",
"mode": "0700"
},
{
@@ -215,164 +270,191 @@
"id": "resolver",
"type": "container",
"name": "mailu-resolver",
"image": "ghcr.io/mailu/unbound@sha256:142aaad82ad1b0d5b59a5f1303778dba61a3e0a540f5d969c48862bcc99f6f5d",
"image": "ghcr.io/mailu/unbound@sha256:3a0fdfb364a63f4f9259526e013c1ef40f5f14de3621ce1560804b3a5909584a",
"network": "mailu",
"env-file": [
"/var/lib/mailu/mailu.env",
"/var/lib/mailu/secret.env"
"${dir:state}/mailu.env",
"${dir:state}/secret.env"
],
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified"
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified",
"ip": "192.168.203.254"
},
{
"id": "redis",
"type": "container",
"name": "mailu-redis",
"image": "redis@sha256:1db42ccef14898aa29bae778452d567534b59c107129cbc1163fb552de184d3c",
"image": "redis@sha256:4bed291aa5efb9f0d77b76ff7d4ab71eee410962965d052552db1fb80576431d",
"network": "mailu",
"volumes": [
"/services/mailu/data/redis:/data"
"${dir:data-redis}:/data"
]
},
{
"id": "admin",
"type": "container",
"name": "mailu-admin",
"image": "ghcr.io/mailu/admin@sha256:dcac20e9cbdad560faef9653b1b5ac0d9266f4098dc00f0e7f0d35f4e70ed8f1",
"image": "ghcr.io/mailu/admin@sha256:6dbfdadc4a9590dcb7652357b505200115b689b74008653bbf369e4599a3be5a",
"network": "mailu",
"env-file": [
"/var/lib/mailu/mailu.env",
"/var/lib/mailu/secret.env",
"/var/lib/mailu/database.env",
"/var/lib/mailu/admin.env"
"${dir:state}/mailu.env",
"${dir:state}/secret.env",
"${dir:state}/database.env",
"${dir:state}/admin.env"
],
"volumes": [
"/services/mailu/data/data:/data",
"/services/mailu/data/dkim:/dkim"
"${dir:data-data}:/data",
"${dir:data-dkim}:/dkim"
],
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified"
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified",
"dns": [
"192.168.203.254"
]
},
{
"id": "imap",
"type": "container",
"name": "mailu-imap",
"image": "ghcr.io/mailu/dovecot@sha256:46d18ba51032be8ebd6841aa49c1ef8762c729038c5fd86a081b5b884d478af9",
"image": "ghcr.io/mailu/dovecot@sha256:7f0ed5db996fbdc00adc5c5e38a08492e04f7eb4a9fbd66a03aa9a28ddf23993",
"network": "mailu",
"env-file": [
"/var/lib/mailu/mailu.env"
"${dir:state}/mailu.env"
],
"volumes": [
"/services/mailu/data/mail:/mail",
"/services/mailu/data/overrides/dovecot:/overrides:ro"
"${dir:data-mail}:/mail",
"${dir:data-overrides-dovecot}:/overrides:ro"
],
"dns": [
"192.168.203.254"
]
},
{
"id": "smtp",
"type": "container",
"name": "mailu-smtp",
"image": "ghcr.io/mailu/postfix@sha256:bbf882880f68849511710b35237a933f3fe80c4b28bf48ff20205dbd1f1433d7",
"image": "ghcr.io/mailu/postfix@sha256:e2e49f39e53b80eac9e7a2f18d9df11edeb4914fd62dbba89b3155e8e034f62e",
"network": "mailu",
"env-file": [
"/var/lib/mailu/mailu.env"
"${dir:state}/mailu.env"
],
"volumes": [
"/services/mailu/data/mailqueue:/queue",
"/services/mailu/data/overrides/postfix:/overrides:ro"
"${dir:data-mailqueue}:/queue",
"${dir:data-overrides-postfix}:/overrides:ro"
],
"dns": [
"192.168.203.254"
]
},
{
"id": "antispam",
"type": "container",
"name": "mailu-antispam",
"image": "ghcr.io/mailu/rspamd@sha256:e87ab93dd252cc69499caa5317dd10d445fd4291a7ecf6bca09793c7d475a0c8",
"image": "ghcr.io/mailu/rspamd@sha256:ff3666d8a61f17d309c5c6f6bcf4d40470b82299ca706ac650301175bb1a079d",
"network": "mailu",
"env-file": [
"/var/lib/mailu/mailu.env"
"${dir:state}/mailu.env"
],
"volumes": [
"/services/mailu/data/filter:/var/lib/rspamd",
"/services/mailu/data/overrides/rspamd:/etc/rspamd/override.d:ro"
"${dir:data-filter}:/var/lib/rspamd",
"${dir:data-overrides-rspamd}:/etc/rspamd/override.d:ro"
],
"dns": [
"192.168.203.254"
]
},
{
"id": "antivirus",
"type": "container",
"name": "mailu-antivirus",
"image": "ghcr.io/mailu/clamav@sha256:01d30483e4a8a20a54566addb1f9b00ebb51e8a103f9226602379c412cf5fb62",
"image": "clamav/clamav-debian@sha256:b12ef8fefddbba7d88de59bea8a32622f365339154adf02d38fd089112e6745a",
"network": "mailu",
"env-file": [
"/var/lib/mailu/mailu.env",
"/var/lib/mailu/secret.env"
],
"volumes": [
"/services/mailu/data/filter:/data"
"${dir:data-clamav}:/var/lib/clamav"
],
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified"
"dns": [
"192.168.203.254"
]
},
{
"id": "webmail",
"type": "container",
"name": "mailu-webmail",
"image": "ghcr.io/mailu/roundcube@sha256:19ccc9c21b2420dabb893ffa707ef90785c785e53dcb6bb9f98da01598412c43",
"image": "ghcr.io/mailu/webmail@sha256:bdbee44cdb05a4658f0e3b62cc448de55ca8f8aea172279fda594826144c04f6",
"network": "mailu",
"env-file": [
"/var/lib/mailu/mailu.env",
"/var/lib/mailu/secret.env"
"${dir:state}/mailu.env",
"${dir:state}/secret.env"
],
"volumes": [
"/services/mailu/data/webmail:/data",
"/services/mailu/data/overrides/roundcube:/overrides:ro"
"${dir:data-webmail}:/data",
"${dir:data-overrides-roundcube}:/overrides:ro"
],
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified"
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified",
"dns": [
"192.168.203.254"
]
},
{
"id": "webdav",
"type": "container",
"name": "mailu-webdav",
"image": "ghcr.io/mailu/radicale@sha256:e13cbad3791c0a6841b5d387e57e49a117808dcef87b8c9969f671ae9c3b67c0",
"image": "ghcr.io/mailu/radicale@sha256:690ed6edf189dfef100a5a8b37c195ebf5d9241ac5f23f2f44b8b7b75726e3de",
"network": "mailu",
"env-file": [
"/var/lib/mailu/mailu.env",
"/var/lib/mailu/secret.env"
"${dir:state}/mailu.env",
"${dir:state}/secret.env"
],
"volumes": [
"/services/mailu/data/dav:/data"
"${dir:data-dav}:/data"
],
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified"
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified",
"dns": [
"192.168.203.254"
]
},
{
"id": "fetchmail",
"type": "container",
"name": "mailu-fetchmail",
"image": "ghcr.io/mailu/fetchmail@sha256:7dcd1392882925d612ab2d0230d437f0c660989d572283c48b0d0f2d491adce7",
"image": "ghcr.io/mailu/fetchmail@sha256:f881c8412d3bbe73d638469b48321558d6403a9d45bfa043c1e52c752103d42d",
"network": "mailu",
"env-file": [
"/var/lib/mailu/mailu.env",
"/var/lib/mailu/secret.env"
"${dir:state}/mailu.env",
"${dir:state}/secret.env"
],
"volumes": [
"/services/mailu/data/data/fetchmail:/data"
"${dir:data-fetchmail}:/data"
],
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified"
"secrets-in-environment": "mailu-admin honours SECRET_KEY_FILE, DB_PW_FILE and API_TOKEN_FILE (configuration.py) but INITIAL_ADMIN_PW is env-only (start.py); the remaining containers' need for SECRET_KEY is unverified",
"dns": [
"192.168.203.254"
]
},
{
"id": "front",
"type": "container",
"name": "mailu-front",
"image": "ghcr.io/mailu/nginx@sha256:09f28ab6d36367fcacc7994f7021f132ac845bdc05f04bf80906102d11aaa057",
"image": "ghcr.io/mailu/nginx@sha256:36f98897cd1bc9d27628bbb4e04bdf60147af2ec7507d6da77f002c4f256896d",
"network": "mailu",
"env-file": [
"/var/lib/mailu/mailu.env"
"${dir:state}/mailu.env"
],
"ports": [
"25",
"110",
"143",
"465",
"587",
"993",
"7080:80"
"995",
"7080:80",
"7443:443"
],
"volumes": [
"/services/mailu/data/certs:/certs",
"/services/mailu/data/overrides/nginx:/overrides:ro"
"${dir:data-certs}:/certs",
"${dir:data-overrides-nginx}:/overrides:ro"
],
"dns": [
"192.168.203.254"
]
},
{
@@ -390,21 +472,40 @@
"network": "mailu",
"volumes": [
"/var/lib/mesh/mailu/broker:/run/secrets/broker:ro",
"/var/lib/mailu/api-token.secret:/run/secrets/api-token:ro",
"${dir:state}/api-token.secret:/run/secrets/api-token:ro",
"${dir:grants}:${dir:grants}:ro",
"/var/lib/mesh/mailu/config.json:/run/config/config.json:ro",
"/var/run/docker.sock:/var/run/docker.sock"
],
"env": {
"MESH_BROKER_FILE": "/run/secrets/broker",
"MESH_MAILU_URL": "http://mailu-admin/api/v1",
"MESH_MAILU_URL": "http://mailu-admin:8080/api/v1",
"MESH_MAILU_API_KEY_FILE": "/run/secrets/api-token",
"MESH_MAILU_IMAP_CONTAINER": "mailu-imap",
"MESH_MAILU_CONFIG_FILE": "/run/config/config.json"
"MESH_MAILU_CONFIG_FILE": "/run/config/config.json",
"MESH_MAILU_DOMAIN": "novox.be",
"MESH_RECEIVES": "${dir:grants}/mesh.json"
},
"restart-on": [
"runtime-config"
],
"artifact": "runtime"
},
{
"id": "automx",
"type": "container",
"name": "mailu-automx",
"artifact": "automx",
"network": "mailu",
"env-file": [
"${dir:state}/mailu.env"
],
"ports": [
"4243"
],
"volumes": [
"${dir:data-automx}:/data"
]
}
],
"build": {
@@ -418,6 +519,10 @@
"arg": "RUNTIME_BASE",
"module": "mesh-tools",
"artifact": "runtime"
},
{
"arg": "PYTHON_BASE",
"image": "python@sha256:25f3cfeaceca14921366af4d1240b56457ef46273bdb508c7b0e8f469f6fd228"
}
],
"artifacts": [
@@ -425,7 +530,31 @@
"name": "runtime",
"kind": "image",
"from": "Dockerfile"
},
{
"name": "automx",
"kind": "image",
"from": "automx/Dockerfile"
}
]
},
"provides": [
{
"name": "smtp",
"scope": "mesh"
}
],
"serves": {
"smtp": {
"port": 587,
"domain": "novox.be",
"name": "mail.novox.be"
}
},
"receives": {
"smtp": "${dir:grants}/mesh.json"
},
"grants": {
"smtp": "${dir:grants}"
}
}
+1 -1
View File
@@ -5,7 +5,7 @@
"type": "module",
"private": true,
"dependencies": {
"@novox/mesh-sdk": "^0.1.0"
"@novox/mesh-sdk": "^0.1.1"
},
"devDependencies": {
"@types/node": "^22.0.0",
+70
View File
@@ -0,0 +1,70 @@
// mailu's provisioner — the adapter that makes mailu a provider of the mesh `smtp` interface.
// The reconcile loop, the contributions file, and reading the mesh's minted password are the sdk
// harness's; this writes only the per-service half: how mailu creates and removes a consumer's
// sending account (novox/hq ADR 0048/0076, gitea's package-registry provisioner is the sibling).
//
// The `smtp` interface: a consumer authenticates to submission (port 587, STARTTLS) as a real
// mailbox this provisioner creates. The address is `<account>@<domain>`: the local part is the
// consumer's `account` contribution — the name it wants to send as — falling back to the mesh's
// own login for a consumer that named none; the domain is the mail server's, which is this
// module's fact, not the consumer's.
//
// **The password is the mesh's, not the provisioner's (ADR 0048).** The mesh mints it and hands
// it to both ends; mailu sets exactly that password every run — so a rotation takes — and seals
// nothing: the consumer already has its copy through the mesh's own channel.
import { runProvisioner, type Provision } from "@novox/mesh-sdk/provisioner";
import { MailuClient } from "../client.js";
const mailu = MailuClient.fromEnv();
// The mail server's own domain. From the environment the manifest composes, because the client's
// config file carries the admin API's coordinates, not the mail domain.
function domain(): string {
const named = (process.env.MESH_MAILU_DOMAIN ?? "").trim();
if (named === "") {
throw new Error("MESH_MAILU_DOMAIN is not set, so a consumer's address cannot be composed");
}
return named;
}
// The address one consumer sends as. The local part is refused rather than sanitised when it is
// not a plain mailbox name — a rewritten name is an address nobody asked for.
function addressOf(p: { as: string; values?: Readonly<Record<string, unknown>> }): string {
const contributed = typeof p.values?.["account"] === "string" ? (p.values["account"] as string).trim() : "";
const local = contributed !== "" ? contributed : p.as;
if (!/^[a-z0-9][a-z0-9._-]*$/.test(local)) {
throw new Error(`${JSON.stringify(local)} is not a usable mailbox name`);
}
return `${local}@${domain()}`;
}
runProvisioner("smtp", {
async create(p: Provision): Promise<void> {
const email = addressOf(p);
// Create if absent, and set exactly the minted password either way so a rotation takes.
// Mailu's create refuses a duplicate address, which is the signal to fall through to the
// password set — the same found-then-apply shape gitea's ensureUser settled on.
try {
await mailu.createUser(email, p.password);
} catch {
await mailu.applyProvisioned(email, p.password);
}
},
async remove(p: { as: string }): Promise<void> {
// The withdrawal only knows the mesh login, never the contributed local part — so accounts
// that contributed one are removed when the address matching the login is absent? No: the
// harness hands remove only `as`, and an address composed from a contribution cannot be
// recomputed from it. The account is therefore removed by its login-shaped address when one
// exists, and left otherwise — a mailbox holding mail is the one thing a background loop
// must not guess about (this module's own events file says the same). Withdrawal of a
// named-account consumer is an operator action until the harness carries values here.
await mailu.deleteUser(`${p.as}@${domain()}`).catch(() => {});
},
// Asked every minute by the harness: whether the backend still holds this consumer exactly as
// the mesh gave it, so a login lost behind the provisioner's back is made again (novox/hq issue 120).
async holds(p: Provision): Promise<boolean> {
return mailu.holdsUser(addressOf(p));
},
});
+1 -1
View File
@@ -7,7 +7,7 @@
],
"claims": [
{
"name": "the-catalogue",
"name": "mesh-catalog",
"scope": "mesh"
}
],
@@ -1,4 +1,4 @@
# verdaccio's runtime: the tool runtime, carrying this module's compiled code.
# minio's runtime: the tool runtime, carrying this module's compiled code.
#
# **Built from this module's own directory and nothing else.** The sdk and the tool runtime are in
# the base images, published like any other artifact — which is what makes this buildable by the
@@ -9,22 +9,32 @@
# image it RUNS in — the second must not carry a compiler. Declared in module.json's `build.on`.
ARG BUILD_BASE
ARG RUNTIME_BASE
ARG MC_CLI
# Named so the final stage's COPY --from can reference a stage, not an ARG — the legacy builder
# this host still runs doesn't expand ARGs inside COPY --from, only inside FROM.
FROM ${MC_CLI} AS mccli
FROM ${BUILD_BASE} AS build
# Compiled under /app/modules so `@novox/mesh-sdk` resolves upward into the base's own
# node_modules — the module is compiled against exactly the sdk it will run against. The compiler
# is invoked by its real path: node_modules/.bin entries are launcher symlinks the base image
# resolved away.
WORKDIR /app/modules/verdaccio
WORKDIR /app/modules/minio
COPY . .
RUN node /app/node_modules/typescript/bin/tsc client.ts index.ts tools/index.ts \
RUN node /app/node_modules/typescript/bin/tsc client.ts tools/index.ts provisioner/index.ts \
--module NodeNext --moduleResolution NodeNext --target ES2022 --outDir dist
FROM ${RUNTIME_BASE}
COPY --from=build /app/modules/verdaccio/dist /app/modules/verdaccio/dist
COPY --from=build /app/modules/minio/dist /app/modules/minio/dist
# The provisioner shells out to mc to actually create buckets and service accounts on the running
# minio server — mc itself was never in this runtime image, only in minio's own. Silently retried
# "spawn mc ENOENT" forever: a requirement was granted at the control-plane level without ever
# materializing the credential on minio. /usr/bin/mc there is a symlink to the real binary, mcli —
# both copied so the symlink resolves.
COPY --from=mccli /usr/bin/mcli /usr/bin/mcli
COPY --from=mccli /usr/bin/mc /usr/bin/mc
# Every serve-time entrypoint, loaded by the runtime in serve mode: tools and events serve, and a
# provider's provisioner runs its reconcile loop in the same process, with the broker connected —
# the convention novox/hq issues 060/061 settled. A container that instead ran only its
# provisioner (`run`) served no tools and emitted no events; a container that named no command
# ran no provisioner at all.
ENV MESH_TOOL_MODULES=/app/modules/verdaccio/dist/index.js,/app/modules/verdaccio/dist/tools/index.js
# the convention novox/hq issues 060/061 settled.
ENV MESH_TOOL_MODULES=/app/modules/minio/dist/tools/index.js,/app/modules/minio/dist/provisioner/index.js
+17 -4
View File
@@ -125,6 +125,18 @@ export class MinioClient {
throw new Error(`minio bucketExists ${bucket}: ${status}`);
}
/**
* Whether a consumer's access key, with exactly this secret, reaches its bucket: a HEAD of the
* bucket signed as the consumer, the way it signs. Read-only. `false` when the key is unknown, the
* secret wrong, access denied or the bucket gone; any other answer rejects (novox/hq issue 120).
*/
async canReachAs(bucket: string, accessKey: string, secretKey: string): Promise<boolean> {
const { status } = await this.request("HEAD", `/${bucket}`, {}, { accessKey, secretKey });
if (status === 200) return true;
if (status === 403 || status === 404) return false;
throw new Error(`minio HEAD ${bucket} as ${accessKey}: ${status}`);
}
async createBucket(bucket: string): Promise<void> {
const { status, text } = await this.request("PUT", `/${bucket}`);
// 200 created; 409 BucketAlreadyOwnedByYou — idempotent, a re-provision must not fail.
@@ -251,6 +263,7 @@ export class MinioClient {
method: string,
path: string,
query: Record<string, string> = {},
as: { accessKey: string; secretKey: string } = { accessKey: this.rootUser, secretKey: this.rootPassword },
): Promise<{ status: number; headers: Headers; text: string }> {
const { amzDate, dateStamp } = this.stamp();
const host = new URL(this.baseUrl).host;
@@ -262,8 +275,8 @@ export class MinioClient {
const canonicalRequest = [method, encodedPath, canonicalQuery, canonicalHeaders, signedHeaders, payloadHash].join("\n");
const scope = `${dateStamp}/${this.region}/s3/aws4_request`;
const stringToSign = ["AWS4-HMAC-SHA256", amzDate, scope, sha256hex(canonicalRequest)].join("\n");
const signature = hmac(this.signingKey(dateStamp), stringToSign).toString("hex");
const authorization = `AWS4-HMAC-SHA256 Credential=${this.rootUser}/${scope}, SignedHeaders=${signedHeaders}, Signature=${signature}`;
const signature = hmac(this.signingKey(dateStamp, as.secretKey), stringToSign).toString("hex");
const authorization = `AWS4-HMAC-SHA256 Credential=${as.accessKey}/${scope}, SignedHeaders=${signedHeaders}, Signature=${signature}`;
const url = `${this.baseUrl}${encodedPath}${canonicalQuery ? `?${canonicalQuery}` : ""}`;
const res = await fetch(url, {
@@ -275,8 +288,8 @@ export class MinioClient {
return { status: res.status, headers: res.headers, text };
}
private signingKey(dateStamp: string): Buffer {
const kDate = hmac(`AWS4${this.rootPassword}`, dateStamp);
private signingKey(dateStamp: string, secretKey: string = this.rootPassword): Buffer {
const kDate = hmac(`AWS4${secretKey}`, dateStamp);
const kRegion = hmac(kDate, this.region);
const kService = hmac(kRegion, "s3");
return hmac(kService, "aws4_request");
+62 -12
View File
@@ -7,6 +7,21 @@
"scope": "mesh"
}
],
"requires": [
"route"
],
"contributes": {
"route": {
"api": {
"label": "files-api",
"port": 9000
},
"console": {
"label": "files",
"port": 9001
}
}
},
"capabilities": [
"container-runtime"
],
@@ -20,12 +35,18 @@
"protocol": "tcp",
"from": "mesh",
"why": "the S3 endpoint"
},
{
"port": 9001,
"protocol": "tcp",
"from": "mesh",
"why": "the admin console"
}
],
"serves": {
"s3-bucket": {
"scheme": "http",
"region": "us-east-1",
"region": "eu-west",
"port": 9000
}
},
@@ -68,20 +89,20 @@
{
"id": "data",
"type": "directory",
"path": "/services/minio/data/data1-1",
"path": "/var/lib/minio-store",
"mode": "0700"
},
{
"id": "net",
"type": "network",
"name": "minio"
"name": "minio-net"
},
{
"id": "server",
"type": "container",
"name": "minio",
"image": "quay.io/minio/minio@sha256:14cea493d9a34af32f524e538b8346cf79f3321eff8e708c1e2960462bd8936e",
"network": "minio",
"image": "docker.io/pgsty/minio@sha256:b6bfe7239bfc83fb90d31612d9704d86039dd714f7904b3f1ad68f211e602372",
"network": "minio-net",
"args": [
"server",
"/data",
@@ -92,22 +113,24 @@
"/var/lib/minio/root.env"
],
"ports": [
"9000"
"9000",
"9001"
],
"volumes": [
"/services/minio/data/data1-1:/data",
"/var/lib/minio-store:/data",
"/var/lib/minio/root.secret:/run/secrets/root:ro"
],
"env": {
"MINIO_ROOT_PASSWORD_FILE": "/run/secrets/root"
"MINIO_ROOT_PASSWORD_FILE": "/run/secrets/root",
"MINIO_BROWSER_REDIRECT_URL": "https://files.novox.be",
"MINIO_REGION": "eu-west"
}
},
{
"id": "runtime",
"type": "container",
"name": "mesh-minio",
"image": "mesh-runtime-minio@sha256:0000000000000000000000000000000000000000000000000000000000000000",
"network": "minio",
"network": "minio-net",
"volumes": [
"/var/lib/mesh/minio/broker:/run/secrets/broker:ro",
"/var/lib/minio/grants:/var/lib/minio/grants:ro",
@@ -117,9 +140,36 @@
"MESH_MINIO_ENDPOINT": "http://minio:9000",
"MESH_MINIO_ROOT_USER": "meshroot",
"MESH_MINIO_ROOT_PASSWORD_FILE": "/run/secrets/root",
"MESH_MINIO_REGION": "eu-west",
"MESH_BROKER_FILE": "/run/secrets/broker",
"MESH_RECEIVES": "/var/lib/minio/grants/mesh.json"
}
},
"artifact": "runtime"
}
]
],
"build": {
"on": [
{
"arg": "BUILD_BASE",
"module": "mesh-tools",
"artifact": "build"
},
{
"arg": "RUNTIME_BASE",
"module": "mesh-tools",
"artifact": "runtime"
},
{
"arg": "MC_CLI",
"image": "docker.io/pgsty/minio@sha256:b6bfe7239bfc83fb90d31612d9704d86039dd714f7904b3f1ad68f211e602372"
}
],
"artifacts": [
{
"name": "runtime",
"kind": "image",
"from": "Dockerfile"
}
]
}
}
+1 -1
View File
@@ -5,7 +5,7 @@
"type": "module",
"private": true,
"dependencies": {
"@novox/mesh-sdk": "^0.1.0"
"@novox/mesh-sdk": "^0.1.1"
},
"devDependencies": {
"@types/node": "^22.0.0",
+6
View File
@@ -53,6 +53,12 @@ runProvisioner("s3-bucket", {
await announce("module.minio.bucket.removed", { bucket, accessKey: p.as });
},
// Asked every minute by the harness: whether the backend still holds this consumer exactly as
// the mesh gave it, so a login lost behind the provisioner's back is made again (novox/hq issue 120).
async holds(p: Provision): Promise<boolean> {
return minio.canReachAs(bucketFor(p.as), p.as, p.password);
},
});
/** Emit best-effort: a broker hiccup is logged and dropped, never allowed to throw back and fail a
+28
View File
@@ -109,6 +109,34 @@ print(EJSON.stringify({ ok: 1 }));
await this.evalJs<{ ok: number }>(js);
}
/**
* Whether `user` authenticates against `database` with exactly `password` and holds `dbOwner`
* there: checked by connecting as the consumer, the way it connects. Read-only. `false` only on an
* authentication failure or a missing role; an unreachable server rejects (novox/hq issue 120).
*/
async canAuthenticateAs(database: string, user: string, password: string): Promise<boolean> {
// Connected without credentials, then authenticated inside the eval from the environment, so
// the consumer's password is neither on argv nor in the message of a failed command.
const uri = `mongodb://${this.conn.host}:${this.conn.port}/?serverSelectionTimeoutMS=10000`;
const js =
"const t = db.getSiblingDB(process.env.MESH_HOLDS_DB);" +
"t.auth(process.env.MESH_HOLDS_USER, process.env.MESH_HOLDS_PW);" +
"print(EJSON.stringify(t.runCommand({ connectionStatus: 1 }).authInfo.authenticatedUserRoles))";
let stdout: string;
try {
({ stdout } = await run("mongosh", [uri, "--quiet", "--eval", js], {
env: { ...process.env, MESH_HOLDS_DB: database, MESH_HOLDS_USER: user, MESH_HOLDS_PW: password },
timeout: 30_000,
}));
} catch (err) {
const text = `${(err as { stderr?: string }).stderr ?? ""}${(err as { stdout?: string }).stdout ?? ""}`;
if (/Authentication failed|AuthenticationFailed/i.test(text)) return false;
throw new Error(`mongosh could not check ${user}: ${text.trim().slice(0, 500) || String((err as Error).message).split("\n")[0]}`);
}
const roles = JSON.parse(stdout.trim()) as { role: string; db: string }[];
return roles.some((r) => r.role === "dbOwner" && r.db === database);
}
/** Drop a database and its owning user, idempotently. Dropping the database evicts its data; the
* user is removed first so a re-grant of the same login starts clean. */
async dropDatabaseAndUser(database: string, user: string): Promise<void> {
+12 -14
View File
@@ -32,13 +32,13 @@
}
},
"receives": {
"mongodb-database": "/var/lib/mongodb/grants/mesh.json"
"mongodb-database": "${dir:grants}/mesh.json"
},
"grants": {
"mongodb-database": "/var/lib/mongodb/grants"
"mongodb-database": "${dir:grants}"
},
"own-secrets": {
"root": "/var/lib/mongodb/root.secret",
"root": "${dir:state}/root.secret",
"broker": "/var/lib/mesh/mongodb/broker"
},
"secrets-owner": "999:999",
@@ -52,19 +52,17 @@
{
"id": "state",
"type": "directory",
"path": "/var/lib/mongodb",
"mode": "0700"
"mode": "0700",
"place": "."
},
{
"id": "grants",
"type": "directory",
"path": "/var/lib/mongodb/grants",
"mode": "0700"
},
{
"id": "data",
"type": "directory",
"path": "/services/mongodb/db-data",
"mode": "0700"
},
{
@@ -75,7 +73,7 @@
{
"id": "server",
"type": "container",
"name": "mongo",
"name": "mongodb-server",
"image": "mongo@sha256:e3fa459b4f4b72f3257c67a23c145e250b8b5700f033860392c68539b998bbe3",
"network": "mongodb",
"env": {
@@ -86,8 +84,8 @@
"27017"
],
"volumes": [
"/services/mongodb/db-data:/data/db",
"/var/lib/mongodb/root.secret:/run/secrets/root:ro"
"${dir:data}:/data/db",
"${dir:state}/root.secret:/run/secrets/root:ro"
]
},
{
@@ -97,14 +95,14 @@
"network": "mongodb",
"volumes": [
"/var/lib/mesh/mongodb/broker:/run/secrets/broker:ro",
"/var/lib/mongodb/grants:/var/lib/mongodb/grants:ro",
"/var/lib/mongodb/root.secret:/run/secrets/root:ro"
"${dir:grants}:${dir:grants}:ro",
"${dir:state}/root.secret:/run/secrets/root:ro"
],
"env": {
"MESH_PROVISION_MONGODB": "mongodb://root@mongo:27017/admin?authSource=admin",
"MESH_PROVISION_MONGODB": "mongodb://root@mongodb-server:27017/admin?authSource=admin",
"MESH_PROVISION_PASSWORD_FILE": "/run/secrets/root",
"MESH_BROKER_FILE": "/run/secrets/broker",
"MESH_RECEIVES": "/var/lib/mongodb/grants/mesh.json"
"MESH_RECEIVES": "${dir:grants}/mesh.json"
},
"artifact": "runtime"
}
+1 -1
View File
@@ -5,7 +5,7 @@
"type": "module",
"private": true,
"dependencies": {
"@novox/mesh-sdk": "^0.1.0"
"@novox/mesh-sdk": "^0.1.1"
},
"devDependencies": {
"@types/node": "^22.0.0",
+5
View File
@@ -45,4 +45,9 @@ runProvisioner("mongodb-database", {
await mongo.dropDatabaseAndUser(p.as, p.as);
await announce("module.mongodb.database.deprovisioned", { database: p.as });
},
// Asked every minute by the harness: whether the backend still holds this consumer exactly as
// the mesh gave it, so a login lost behind the provisioner's back is made again (novox/hq issue 120).
async holds(p: Provision): Promise<boolean> {
return mongo.canAuthenticateAs(p.as, p.as, p.password);
},
});
+94 -3
View File
@@ -15,6 +15,7 @@
// The one cost dynsec carries is the bootstrap file; see initBootstrapFile() and the module README.
import { randomBytes } from "node:crypto";
import { connect as tcpConnect } from "node:net";
import { readFileSync } from "node:fs";
import { execFile } from "node:child_process";
import { promisify } from "node:util";
@@ -87,9 +88,20 @@ export class MosquittoClient {
"-u", this.conn.adminUser,
"-P", this.conn.adminPassword,
];
const { stdout, stderr } = await run("mosquitto_ctrl", [...base, "dynsec", ...args], {
maxBuffer: 16 << 20,
});
let stdout: string;
let stderr: string;
try {
({ stdout, stderr } = await run("mosquitto_ctrl", [...base, "dynsec", ...args], {
maxBuffer: 16 << 20,
timeout: 30_000,
}));
} catch (err) {
// A failed run's message repeats its argv, the admin password (-P) included; say what failed
// without it.
const e = err as { code?: unknown; signal?: unknown; stderr?: string; stdout?: string };
const detail = `${e.stderr ?? ""}${e.stdout ?? ""}`.trim().slice(0, 500);
throw new Error(`mosquitto_ctrl dynsec ${args[0] ?? ""} could not run (${e.code ?? e.signal ?? "error"}): ${detail}`);
}
const failure = ctlError(`${stdout}\n${stderr}`);
if (failure) {
throw new Error(`mosquitto_ctrl dynsec ${args[0] ?? ""} failed: ${failure}`);
@@ -140,6 +152,11 @@ export class MosquittoClient {
if (await this.clientExists(username)) {
await this.ctl("setClientPassword", username, password);
// A disabled client is refused like a wrong password, so the check the provisioner runs
// reports it lost; applying again must enable it, or the two would disagree for ever.
if (/Disabled:\s*true/i.test(await this.ctl("getClient", username))) {
await this.ctl("enableClient", username);
}
} else {
await this.ctl("createClient", username, "-p", password);
}
@@ -163,6 +180,28 @@ export class MosquittoClient {
}
}
/**
* Whether a consumer's client accepts exactly this password and still carries its own role.
* Read-only. The password is checked the way the consumer is checked, by an MQTT CONNECT as it,
* and the broker's CONNACK code is the answer: 0 accepted, 4 bad credentials, 5 not authorised.
* Nothing rides on argv. An unreachable broker rejects (novox/hq issue 120).
*/
async holdsClient(username: string, password: string): Promise<boolean> {
const code = await mqttConnack(this.conn.host, this.conn.port, username, password);
if (code === 4 || code === 5) return false;
if (code !== 0) throw new Error(`mosquitto refused ${username} with CONNACK ${code}`);
// The role, asked directly: only "not found" means absent. Any other failure to ask rejects,
// unlike clientHasRole, which reads every failure as "no role".
let out: string;
try {
out = await this.ctl("getClient", username);
} catch (err) {
if (/not\s*found|does not exist|no such/i.test(String(err))) return false;
throw err;
}
return new RegExp(`(^|\\s)${escapeRegExp(username)}\\s+\\(priority`, "m").test(out);
}
/** Remove a client and the per-client role created for it, idempotently. */
async deleteScopedClient(username: string): Promise<void> {
await ignoreMissing(this.ctl("deleteClient", username));
@@ -249,3 +288,55 @@ function readSecretFile(path: string | undefined): string | undefined {
return undefined;
}
}
/**
* Connect once over MQTT 3.1.1 with a username and password, return the broker's CONNACK return code,
* and disconnect. A clean session under a throwaway client id, so no consumer session is taken over.
*/
function mqttConnack(host: string, port: number, username: string, password: string): Promise<number> {
const str = (v: string): Buffer => {
const b = Buffer.from(v, "utf8");
const len = Buffer.alloc(2);
len.writeUInt16BE(b.length);
return Buffer.concat([len, b]);
};
const variable = Buffer.concat([str("MQTT"), Buffer.from([4, 0xc2, 0, 10])]); // level 4; user+pass+clean; keepalive 10s
const payload = Buffer.concat([str(`mesh-holds-${randomBytes(6).toString("hex")}`), str(username), str(password)]);
let remaining = variable.length + payload.length;
const lenBytes: number[] = [];
do {
let byte = remaining % 128;
remaining = Math.floor(remaining / 128);
if (remaining > 0) byte |= 0x80;
lenBytes.push(byte);
} while (remaining > 0);
const packet = Buffer.concat([Buffer.from([0x10, ...lenBytes]), variable, payload]);
return new Promise((resolve, reject) => {
const socket = tcpConnect({ host, port });
let buf = Buffer.alloc(0);
const timer = setTimeout(() => {
socket.destroy();
reject(new Error(`no CONNACK from ${host}:${port} within 10s`));
}, 10_000);
socket.on("connect", () => socket.write(packet));
socket.on("data", (chunk) => {
buf = Buffer.concat([buf, chunk]);
if (buf.length < 4) return;
clearTimeout(timer);
if (buf[0] !== 0x20) {
socket.destroy();
reject(new Error(`unexpected MQTT packet 0x${buf[0].toString(16)} instead of CONNACK`));
return;
}
const code = buf[3];
if (code === 0) socket.end(Buffer.from([0xe0, 0])); // DISCONNECT
else socket.destroy();
resolve(code);
});
socket.on("error", (err) => {
clearTimeout(timer);
reject(err);
});
});
}
+1 -1
View File
@@ -5,7 +5,7 @@
"type": "module",
"private": true,
"dependencies": {
"@novox/mesh-sdk": "^0.1.0"
"@novox/mesh-sdk": "^0.1.1"
},
"devDependencies": {
"@types/node": "^22.0.0",
+5
View File
@@ -43,4 +43,9 @@ runProvisioner("mqtt-topic", {
await mosquitto.deleteScopedClient(p.as);
await announce("module.mosquitto.topic.deprovisioned", { username: p.as });
},
// Asked every minute by the harness: whether the backend still holds this consumer exactly as
// the mesh gave it, so a login lost behind the provisioner's back is made again (novox/hq issue 120).
async holds(p: Provision): Promise<boolean> {
return mosquitto.holdsClient(p.as, p.password);
},
});
+54 -4
View File
@@ -75,14 +75,18 @@ export class MssqlClient {
* prints (split across output lines for a large result, and reassembled here) is parsed. An
* empty result yields no output at all — an empty array.
*/
async query(select: string, database = "master"): Promise<Record<string, unknown>[]> {
async query(
select: string,
database = "master",
variables: Record<string, string> = {},
): Promise<Record<string, unknown>[]> {
const wrapped = `SET NOCOUNT ON;\n${stripTrailingSemis(select)}\nFOR JSON PATH, INCLUDE_NULL_VALUES;`;
const stdout = await this.sqlcmd(wrapped, database);
const stdout = await this.sqlcmd(wrapped, database, variables);
return parseJsonRows(stdout);
}
/** The one execution boundary: invoke `sqlcmd` and return its concatenated stdout. */
private async sqlcmd(sql: string, database: string): Promise<string> {
private async sqlcmd(sql: string, database: string, variables: Record<string, string> = {}): Promise<string> {
// `-h -1` drops the column-header rule; `-y 0`/`-Y 0` lift the display-width cap so a long
// JSON document is not truncated; `-W` trims trailing whitespace so the JSON chunks rejoin
// cleanly. sqlcmd from the mssql-tools ships in the runtime container, the way `psql` ships
@@ -101,7 +105,9 @@ export class MssqlClient {
"-W",
"-Q", sql,
],
{ env: { ...process.env, SQLCMDPASSWORD: this.conn.password }, maxBuffer: 16 << 20 },
// `variables` reach sqlcmd as environment variables, which it substitutes as `$(NAME)` scripting
// variables: a value that must not appear on argv, or in the message of a failed command.
{ env: { ...process.env, ...variables, SQLCMDPASSWORD: this.conn.password }, maxBuffer: 16 << 20 },
);
return stdout;
}
@@ -121,6 +127,9 @@ export class MssqlClient {
);
} else {
await this.exec(`ALTER LOGIN ${ident(login)} WITH PASSWORD = ${literal(password)}`);
// A disabled login is refused like a wrong password; the check the provisioner runs reports it
// lost, so applying again must enable it or the two would disagree for ever.
await this.exec(`ALTER LOGIN ${ident(login)} ENABLE`);
}
const dbs = await this.query(
@@ -138,10 +147,51 @@ export class MssqlClient {
);
if (users.length === 0) {
await this.exec(`CREATE USER ${ident(login)} FOR LOGIN ${ident(login)}`, database);
} else {
// Re-point an existing user at the login when its SID is not the login's: a database restored
// from elsewhere keeps its user under the old login's SID, orphaned. Only then, so a user that
// is already mapped is left alone.
const orphaned = await this.query(
`SELECT 1 AS ok FROM sys.database_principals WHERE name = ${literal(login)} ` +
`AND (sid IS NULL OR sid <> SUSER_SID(${literal(login)}))`,
database,
);
if (orphaned.length > 0) {
await this.exec(`ALTER USER ${ident(login)} WITH LOGIN = ${ident(login)}`, database);
}
}
await this.exec(`ALTER ROLE db_owner ADD MEMBER ${ident(login)}`, database);
}
/**
* Whether `login` exists, is enabled, has exactly `password`, and is a db_owner user of
* `database`. Read-only: the password is compared with PWDCOMPARE against the stored hash, so
* nothing logs in and no failed-login is recorded (novox/hq issue 120).
*/
async holdsLogin(database: string, login: string, password: string): Promise<boolean> {
// The password reaches sqlcmd as a scripting variable from the environment, never inside the
// query text, so it is neither on argv nor in the message of a failed command. It is the mesh's
// minted value, which carries no quote.
const server = await this.query(
`SELECT CAST(CASE WHEN EXISTS (SELECT 1 FROM sys.sql_logins WHERE name = ${literal(login)} ` +
`AND is_disabled = 0 AND PWDCOMPARE(N'$(MESHHOLDSPW)', password_hash) = 1) ` +
`AND DB_ID(${literal(database)}) IS NOT NULL THEN 1 ELSE 0 END AS int) AS ok`,
"master",
{ MESHHOLDSPW: password },
);
if (Number(server[0]?.ok) !== 1) return false;
// The user must be this login's, by SID, and a db_owner. A user orphaned by a restore has the
// right name and the wrong SID, and cannot be reached through the login.
const owner = await this.query(
`SELECT CAST(CASE WHEN EXISTS (SELECT 1 FROM sys.database_principals dp ` +
`JOIN sys.server_principals sp ON dp.sid = sp.sid ` +
`WHERE dp.name = ${literal(login)} AND sp.name = ${literal(login)}) ` +
`AND IS_ROLEMEMBER('db_owner', ${literal(login)}) = 1 THEN 1 ELSE 0 END AS int) AS ok`,
database,
);
return Number(owner[0]?.ok) === 1;
}
/** Drop a database and its login, idempotently, after evicting live connections. */
async dropDatabaseAndLogin(database: string, login: string): Promise<void> {
const dbs = await this.query(
+2 -3
View File
@@ -68,7 +68,6 @@
{
"id": "data",
"type": "directory",
"path": "/services/mssql/data",
"mode": "0700",
"owner": "10001:0"
},
@@ -87,10 +86,10 @@
"/var/lib/mssql/sa.env"
],
"ports": [
"4848:1433"
"1433"
],
"volumes": [
"/services/mssql/data:/var/opt/mssql"
"${dir:data}:/var/opt/mssql"
],
"secrets-in-environment": "the image documents only MSSQL_SA_PASSWORD, no _FILE and no configuration field; not convertible without a wrapper entrypoint"
},
+1 -1
View File
@@ -5,7 +5,7 @@
"type": "module",
"private": true,
"dependencies": {
"@novox/mesh-sdk": "^0.1.0"
"@novox/mesh-sdk": "^0.1.1"
},
"devDependencies": {
"@types/node": "^22.0.0",
+5
View File
@@ -44,4 +44,9 @@ runProvisioner("mssql-database", {
await mssql.dropDatabaseAndLogin(p.as, p.as);
await announce("module.mssql.database.deprovisioned", { database: p.as });
},
// Asked every minute by the harness: whether the backend still holds this consumer exactly as
// the mesh gave it, so a login lost behind the provisioner's back is made again (novox/hq issue 120).
async holds(p: Provision): Promise<boolean> {
return mssql.holdsLogin(p.as, p.as, p.password);
},
});
+1 -1
View File
@@ -71,7 +71,7 @@
"/var/lib/n8n/server.env"
],
"ports": [
"5682:5678"
"5678"
],
"volumes": [
"/services/n8n/n8n-data:/home/node/.n8n"
+36
View File
@@ -0,0 +1,36 @@
{
"module": "networkmanager",
"version": "1",
"capabilities": [
"package-manager",
"service-manager"
],
"claims": [
{
"name": "node-uplink",
"scope": "node"
}
],
"resources": [
{
"id": "package",
"type": "package",
"package": "networkmanager"
},
{
"id": "config",
"type": "file",
"path": "/etc/NetworkManager/conf.d/50-mesh.conf",
"mode": "0644",
"content": "# Managed by the mesh (module networkmanager). Replaced on every push; edit the\n# catalogue instead.\n#\n# This machine's uplink is NetworkManager's, and this file is the whole of what\n# the mesh asks of it (novox/hq ADR 0117): leave the resolver file to the mesh,\n# and leave the private network's interface alone. Nothing more. The mesh never\n# declares a connection profile, an address, a route, a wireless network or its\n# credentials \u2014 those are joined at the machine, by the person using it, and\n# the link they make is the only channel the mesh reaches this machine over. A\n# push that got a link wrong could not be undone by the next one.\n#\n# A drop-in of the mesh's own, beside NetworkManager.conf and whatever else the\n# operator keeps in this directory. NetworkManager reads the files here sorted by\n# name and a later one wins a key it sets again \u2014 so a file of the operator's\n# that sorts after this one (any name starting with a letter does) and sets dns=\n# or unmanaged-devices= overrides it. That is the operator's to decide, and the\n# reason this file sets nothing but the two keys it must.\n#\n# NetworkManager itself is the machine's: the mesh never starts, stops, enables\n# or disables it (its service is declared with no state), because stopping it\n# takes every link down, this machine's channel to the mesh included \u2014 and a\n# module unassigned by mistake must not be able to do that. When this file\n# changes, a running NetworkManager is reloaded (its D-Bus Reload call, which\n# re-reads its configuration \u2014 NetworkManager(8)), never restarted.\n\n[main]\n# The resolver file is the mesh's: resolv-conf writes /etc/resolv.conf and names\n# the mesh's resolver. Without this line NetworkManager rewrites that file on\n# every connectivity change \u2014 every network joined, every lease renewed \u2014\n# and the mesh's resolver is silently replaced while every surface of the mesh\n# still reads green. none: \"NetworkManager will not modify resolv.conf. This\n# implies rc-manager unmanaged\" (NetworkManager.conf(5), 1.58). On an adopted\n# machine the predecessor wrote the same line in a file of its own; both say one\n# thing, and the predecessor's is retired by hand after the take.\ndns=none\n\n[keyfile]\n# mesh0 is the private network's interface: the mesh brings it up and the mesh\n# alone configures it. A manager that considers every interface its own could\n# try to configure it, or tear it down on a profile change.\n#\n# unmanaged-devices rather than a [device-mesh0] section with managed=0, because\n# NetworkManager.conf(5) says a device unmanaged by this key \"is strictly\n# unmanaged and cannot be overruled by using the API like nmcli device set\n# $IFNAME managed yes\", while device*.managed \"can be overruled at runtime via\n# D-Bus\". The same page adds that device*.managed \"may be a better choice\" for\n# exactly those reasons \u2014 for an interface the operator might want to hand back\n# at runtime. For the mesh's own interface, strict is the point.\n#\n# += rather than =: the same page documents appending to a list-valued key set\n# earlier (\"plugins+=another-plugin\") as an extension of its key file format,\n# and unmanaged-devices is a device list. = would replace whatever devices the\n# operator already keeps NetworkManager away from; += adds this one to them\n# (novox/hq ADR 0102: a list is added to, never replaced). A file of the\n# operator's read after this one that sets the key with = replaces it again;\n# that is the operator's to decide.\nunmanaged-devices+=interface-name:mesh0\n"
},
{
"id": "service",
"type": "service",
"unit": "NetworkManager.service",
"reload-on": [
"config"
]
}
]
}
+10
View File
@@ -9,6 +9,11 @@
# image it RUNS in — the second must not carry a compiler. Declared in module.json's `build.on`.
ARG BUILD_BASE
ARG RUNTIME_BASE
ARG DOCKER_CLI
# Named so the final stage's COPY --from can reference a stage, not an ARG — the legacy builder
# this host still runs doesn't expand ARGs inside COPY --from, only inside FROM.
FROM ${DOCKER_CLI} AS dockercli
FROM ${BUILD_BASE} AS build
# Compiled under /app/modules so `@novox/mesh-sdk` resolves upward into the base's own
@@ -22,6 +27,11 @@ RUN node /app/node_modules/typescript/bin/tsc client.ts index.ts tools/index.ts
FROM ${RUNTIME_BASE}
COPY --from=build /app/modules/nextcloud/dist /app/modules/nextcloud/dist
# occ runs inside nextcloud's own container, reached over the mounted docker socket — which needs
# the docker CLI itself present here, not only the socket. Copied from Docker's own official client
# image rather than apt-installed, so this stays the one binary and nothing else (no daemon, no
# systemd unit, no package manager tree pulled in for it).
COPY --from=dockercli /usr/local/bin/docker /usr/local/bin/docker
# Every serve-time entrypoint, loaded by the runtime in serve mode: tools and events serve, and a
# provider's provisioner runs its reconcile loop in the same process, with the broker connected —
# the convention novox/hq issues 060/061 settled. A container that instead ran only its
+7 -2
View File
@@ -52,8 +52,13 @@ export class NextcloudClient {
const container = cfg.container ?? env.MESH_NEXTCLOUD_CONTAINER ?? "nextcloud";
const ocsUrl = cfg.url ?? env.MESH_NEXTCLOUD_URL ?? `http://127.0.0.1:${env.NEXTCLOUD_PORT ?? "80"}`;
const adminUser = cfg.user ?? env.MESH_NEXTCLOUD_ADMIN_USER ?? "admin";
const adminPassword = cfg.password ?? env.MESH_NEXTCLOUD_ADMIN_PASSWORD;
if (!adminPassword) throw new Error("no Nextcloud admin password — set MESH_NEXTCLOUD_ADMIN_PASSWORD");
const passwordFile = env.MESH_NEXTCLOUD_ADMIN_PASSWORD_FILE;
const adminPassword = cfg.password ?? env.MESH_NEXTCLOUD_ADMIN_PASSWORD
?? (passwordFile ? readFileSync(passwordFile, "utf8").trim() : undefined);
if (!adminPassword) {
throw new Error("no Nextcloud admin password — set MESH_NEXTCLOUD_ADMIN_PASSWORD_FILE " +
"(or MESH_NEXTCLOUD_ADMIN_PASSWORD)");
}
return new NextcloudClient(container, ocsUrl.replace(/\/$/, ""), adminUser, adminPassword);
}
+22 -19
View File
@@ -11,29 +11,26 @@
"postgres-database": {
"name": "nextcloud"
},
"s3-bucket": {
"bucket": "nextcloud"
},
"route": {
"label": "drive",
"port": 80
}
},
"binds": {
"postgres-database": "/var/lib/nextcloud-module/database.json",
"s3-bucket": "/var/lib/nextcloud-module/store.json",
"route": "/var/lib/nextcloud-module/route.json"
"postgres-database": "${dir:state}/database.json",
"s3-bucket": "${dir:state}/store.json",
"route": "${dir:state}/route.json"
},
"secrets": {
"postgres-database": "/var/lib/nextcloud-module/database.secret",
"s3-bucket": "/var/lib/nextcloud-module/store.secret"
"postgres-database": "${dir:state}/database.secret",
"s3-bucket": "${dir:state}/store.secret"
},
"emits": [
"module.nextcloud.user.created",
"module.nextcloud.share.created"
],
"own-secrets": {
"admin": "/var/lib/nextcloud-module/admin.secret",
"admin": "${dir:state}/admin.secret",
"broker": "/var/lib/mesh/nextcloud/broker"
},
"capabilities": [
@@ -57,20 +54,19 @@
{
"id": "state",
"type": "directory",
"path": "/var/lib/nextcloud-module",
"mode": "0700"
"mode": "0700",
"place": "."
},
{
"id": "server-env",
"type": "file",
"path": "/var/lib/nextcloud-module/server.env",
"path": "${dir:state}/server.env",
"mode": "0600",
"content": "POSTGRES_HOST=${bound:postgres-database:at}:${bound:postgres-database:port}\nPOSTGRES_DB=${bound:postgres-database:as}\nPOSTGRES_USER=${bound:postgres-database:as}\nPOSTGRES_PASSWORD=${secret:postgres-database}\nNEXTCLOUD_ADMIN_USER=admin\nNEXTCLOUD_ADMIN_PASSWORD=${secret:admin}\nOBJECTSTORE_S3_HOST=${bound:s3-bucket:at}\nOBJECTSTORE_S3_PORT=${bound:s3-bucket:port}\nOBJECTSTORE_S3_BUCKET=nextcloud\nOBJECTSTORE_S3_KEY=${bound:s3-bucket:as}\nOBJECTSTORE_S3_SECRET=${secret:s3-bucket}\nOBJECTSTORE_S3_SSL=false\nOBJECTSTORE_S3_USEPATH_STYLE=true\n"
"content": "POSTGRES_HOST=${bound:postgres-database:at}:${bound:postgres-database:port}\nPOSTGRES_DB=${bound:postgres-database:as}\nPOSTGRES_USER=${bound:postgres-database:as}\nPOSTGRES_PASSWORD=${secret:postgres-database}\nNEXTCLOUD_ADMIN_USER=mesh-admin\nNEXTCLOUD_ADMIN_PASSWORD=${secret:admin}\nOBJECTSTORE_S3_HOST=${bound:s3-bucket:at}\nOBJECTSTORE_S3_PORT=${bound:s3-bucket:port}\nOBJECTSTORE_S3_BUCKET=mesh-novox-ncloud\nOBJECTSTORE_S3_KEY=${bound:s3-bucket:as}\nOBJECTSTORE_S3_SECRET=${secret:s3-bucket}\nOBJECTSTORE_S3_SSL=false\nOBJECTSTORE_S3_USEPATH_STYLE=true\nOBJECTSTORE_S3_REGION=${bound:s3-bucket:region}\n"
},
{
"id": "html",
"type": "directory",
"path": "/services/nextcloud/html",
"mode": "0750",
"owner": "33:33"
},
@@ -78,15 +74,15 @@
"id": "server",
"type": "container",
"name": "nextcloud",
"image": "nextcloud@sha256:0b8261f6335af6b95264ce893b4d645857638e0fa151b5ba620f25f377318ae1",
"image": "nextcloud@sha256:fb966733647ea03f0446b0c22eac9733c8eb616d37b960caca9d4c3010e14a08",
"env-file": [
"/var/lib/nextcloud-module/server.env"
"${dir:state}/server.env"
],
"ports": [
"80"
],
"volumes": [
"/services/nextcloud/html:/var/www/html"
"${dir:html}:/var/www/html"
],
"secrets-in-environment": "the image honours POSTGRES_PASSWORD_FILE and NEXTCLOUD_ADMIN_PASSWORD_FILE (entrypoint file_env); OBJECTSTORE_S3_SECRET has none and needs a generated config fragment; convertible, awaiting a bed"
},
@@ -106,12 +102,15 @@
"volumes": [
"/var/lib/mesh/nextcloud/broker:/run/secrets/broker:ro",
"/var/lib/mesh/nextcloud/config.json:/run/config/config.json:ro",
"${dir:state}/admin.secret:/run/secrets/admin:ro",
"/var/run/docker.sock:/var/run/docker.sock"
],
"env": {
"MESH_BROKER_FILE": "/run/secrets/broker",
"MESH_NEXTCLOUD_URL": "http://127.0.0.1:80",
"MESH_NEXTCLOUD_CONFIG_FILE": "/run/config/config.json"
"MESH_NEXTCLOUD_URL": "http://127.0.0.1:${port:80}",
"MESH_NEXTCLOUD_CONFIG_FILE": "/run/config/config.json",
"MESH_NEXTCLOUD_ADMIN_USER": "mesh-admin",
"MESH_NEXTCLOUD_ADMIN_PASSWORD_FILE": "/run/secrets/admin"
},
"restart-on": [
"runtime-config"
@@ -130,6 +129,10 @@
"arg": "RUNTIME_BASE",
"module": "mesh-tools",
"artifact": "runtime"
},
{
"arg": "DOCKER_CLI",
"image": "docker@sha256:018edbc908e08fcc9dbf029c812c34251e9b4719e6f71ca0e5eae2a987d014ca"
}
],
"artifacts": [
+2 -2
View File
@@ -6,7 +6,7 @@
],
"claims": [
{
"name": "the-packet-filter",
"name": "node-packet-filter",
"scope": "node"
}
],
@@ -30,7 +30,7 @@
"id": "stock-unit-stop",
"type": "file",
"path": "/etc/systemd/system/nftables.service.d/mesh.conf",
"content": "# The mesh: stopping the stock unit deletes only the mesh's table, never the whole ruleset\n# (novox/hq ADR 0100) — a flush would take the container runtime's rules and any firewall with it.\n[Service]\nExecStop=\nExecStop=nft delete table inet mesh\n",
"content": "# The mesh: stopping the stock unit deletes only the mesh's table, never the whole ruleset\n# (novox/hq ADR 0100) \u2014 a flush would take the container runtime's rules and any firewall with it.\n[Service]\nExecStop=\nExecStop=nft delete table inet mesh\n",
"mode": "0644"
},
{
+1 -1
View File
@@ -43,7 +43,7 @@
"image": "registry-api.novox.be/novox/www@sha256:aa7ed20a293e1d7444c5c5d59b6d8bdb382bad159559a22e006810e379cae69c",
"network": "novox-be",
"ports": [
"4000:8080"
"8080"
]
}
]
+13 -20
View File
@@ -15,10 +15,10 @@
}
},
"binds": {
"route": "/var/lib/only-office/route.json"
"route": "${dir:state}/route.json"
},
"own-secrets": {
"jwt": "/var/lib/only-office/jwt.secret"
"jwt": "${dir:state}/jwt.secret"
},
"listens": [
{
@@ -32,56 +32,49 @@
{
"id": "state",
"type": "directory",
"path": "/var/lib/only-office",
"mode": "0700"
"mode": "0700",
"place": "."
},
{
"id": "server-env",
"type": "file",
"path": "/var/lib/only-office/server.env",
"path": "${dir:state}/server.env",
"mode": "0600",
"content": "JWT_ENABLED=true\nJWT_SECRET=${secret:jwt}\nJWT_HEADER=Authorization\nJWT_IN_BODY=true\nALLOW_PRIVATE_IP_ADDRESS=true\n"
},
{
"id": "logs",
"type": "directory",
"path": "/services/only-office/logs",
"mode": "0700"
},
{
"id": "data",
"type": "directory",
"path": "/services/only-office/data",
"mode": "0700"
},
{
"id": "lib",
"type": "directory",
"path": "/services/only-office/lib",
"mode": "0700"
},
{
"id": "db",
"type": "directory",
"path": "/services/only-office/db",
"mode": "0700"
},
{
"id": "rabbitmq",
"type": "directory",
"path": "/services/only-office/rabbitmq",
"mode": "0700"
},
{
"id": "redis",
"type": "directory",
"path": "/services/only-office/redis",
"mode": "0700"
},
{
"id": "fonts",
"type": "directory",
"path": "/services/only-office/fonts",
"mode": "0700"
},
{
@@ -96,19 +89,19 @@
"image": "onlyoffice/documentserver@sha256:3ab6ebc7c605e5a32b7ae3ff19daed4925090245acc8100ce2230bd766c88212",
"network": "only-office",
"env-file": [
"/var/lib/only-office/server.env"
"${dir:state}/server.env"
],
"ports": [
"9070:80"
],
"volumes": [
"/services/only-office/logs:/var/log/onlyoffice",
"/services/only-office/data:/var/www/onlyoffice/Data",
"/services/only-office/lib:/var/lib/onlyoffice",
"/services/only-office/db:/var/lib/postgresql",
"/services/only-office/rabbitmq:/var/lib/rabbitmq",
"/services/only-office/redis:/var/lib/redis",
"/services/only-office/fonts:/usr/share/fonts/truetype/custom"
"${dir:logs}:/var/log/onlyoffice",
"${dir:data}:/var/www/onlyoffice/Data",
"${dir:lib}:/var/lib/onlyoffice",
"${dir:db}:/var/lib/postgresql",
"${dir:rabbitmq}:/var/lib/rabbitmq",
"${dir:redis}:/var/lib/redis",
"${dir:fonts}:/usr/share/fonts/truetype/custom"
],
"secrets-in-environment": "run-document-server.sh regenerates JWT_SECRET from the environment on every start and overwrites local.json; not convertible"
}
+1 -1
View File
@@ -44,7 +44,7 @@
"image": "registry-api.novox.be/novox/photos-client@sha256:f87d63ee7bfb44c9f9748b99be6ba6dc0daf955a1d463699e9e7e3009693c0ab",
"network": "photos-eef",
"ports": [
"4012:80"
"80"
]
}
]
+1 -1
View File
@@ -44,7 +44,7 @@
"image": "registry-api.novox.be/novox/photos-client@sha256:f87d63ee7bfb44c9f9748b99be6ba6dc0daf955a1d463699e9e7e3009693c0ab",
"network": "photos-filip",
"ports": [
"4013:80"
"80"
]
}
]
+2 -5
View File
@@ -10,9 +10,6 @@
"route"
],
"contributes": {
"s3-bucket": {
"bucket": "photos"
},
"mongodb-database": {
"name": "photos"
},
@@ -56,7 +53,7 @@
"type": "file",
"path": "/var/lib/photos/server.env",
"mode": "0600",
"content": "NODE_ENV=production\nPORT=9000\nMONGO_URL=mongodb://${bound:mongodb-database:as}:${secret:mongodb-database}@${bound:mongodb-database:at}:${bound:mongodb-database:port}/${bound:mongodb-database:as}?authSource=admin\nMONGO_DB=${bound:mongodb-database:as}\nMINIO_ENDPOINT=${bound:s3-bucket:at}\nMINIO_PORT=${bound:s3-bucket:port}\nMINIO_BUCKET=photos\nMINIO_ACCESSKEY=${bound:s3-bucket:as}\nMINIO_SECRET=${secret:s3-bucket}\nMINIO_USE_SSL=false\n"
"content": "NODE_ENV=production\nPORT=9000\nMONGO_URL=mongodb://${bound:mongodb-database:as}:${secret:mongodb-database}@${bound:mongodb-database:at}:${bound:mongodb-database:port}/${bound:mongodb-database:as}?authSource=admin\nMONGO_DB=${bound:mongodb-database:as}\nMINIO_ENDPOINT=${bound:s3-bucket:at}\nMINIO_PORT=${bound:s3-bucket:port}\nMINIO_BUCKET=mesh-novox-photos\nMINIO_ACCESSKEY=${bound:s3-bucket:as}\nMINIO_SECRET=${secret:s3-bucket}\nMINIO_USE_SSL=false\n"
},
{
"id": "net",
@@ -84,7 +81,7 @@
"image": "registry-api.novox.be/novox/photos-admin-client@sha256:f437fa9ed28b29a012f715fb8d9b809a15cff4a672794c620d5d400f57695580",
"network": "photos",
"ports": [
"4001:80"
"80"
]
}
]
+23 -5
View File
@@ -6,11 +6,17 @@
"container-runtime"
],
"listens": [
{
"port": 9090,
"protocol": "tcp",
"from": "mesh",
"why": "the dashboard over http; portainer.novox.be is a route grant and the proxy reaches it here \u2014 the machine side of 9090:9000, the predecessor's number"
},
{
"port": 9443,
"protocol": "tcp",
"from": "mesh",
"why": "the container dashboard, over its own tls"
"why": "the same dashboard over its own tls; the runtime sidecar talks to it here"
}
],
"resources": [
@@ -23,19 +29,19 @@
{
"id": "data",
"type": "directory",
"path": "/services/portainer/data",
"mode": "0700"
},
{
"id": "server",
"type": "container",
"name": "portainer",
"image": "portainer/portainer-ce@sha256:511f3f06c96fe3b993ebeaafde311c1959cae73a7ef825dba6397d51b450dffa",
"image": "portainer/portainer-ce@sha256:4d616db18cfeb5dd41a69c0958bc825c84483ea9cde1106eb82a5d26f3bd8b0e",
"ports": [
"9443"
"9090:9000",
"9443:9443"
],
"volumes": [
"/services/portainer/data:/data",
"${dir:data}:/data",
"/var/run/docker.sock:/var/run/docker.sock"
]
},
@@ -90,5 +96,17 @@
"from": "Dockerfile"
}
]
},
"requires": [
"route"
],
"contributes": {
"route": {
"label": "portainer",
"port": 9090
}
},
"binds": {
"route": "/var/lib/mesh/portainer/route.json"
}
}
+38 -3
View File
@@ -39,7 +39,16 @@ export class PostgresClient {
static fromEnv(env: NodeJS.ProcessEnv = process.env): PostgresClient {
const url = env.MESH_PROVISION_POSTGRES ? safeUrl(env.MESH_PROVISION_POSTGRES) : undefined;
const host = env.MESH_POSTGRES_HOST ?? url?.hostname;
const port = Number(env.MESH_POSTGRES_PORT ?? url?.port ?? "5432") || 5432;
// MESH_PROVISION_POSTGRES_PORT is the seat's twin (mesh-controller's own
// internal/envfile.Placed pattern): which port this machine actually put mesh-store at, when
// that differs from what the connection string above already says — e.g. adopted in place at
// a predecessor's port. Empty means the mesh has nothing to add and the string's own port
// stands; a placeholder the mesh never filled (a manifest ahead of the running controller)
// is treated the same way, not as a fault.
const seatPort = (env.MESH_PROVISION_POSTGRES_PORT ?? "").trim();
const filledSeatPort = seatPort && !/^\$\{[^}]*\}$/.test(seatPort) ? seatPort : undefined;
const portSource = env.MESH_POSTGRES_PORT ?? filledSeatPort ?? url?.port ?? "5432";
const port = Number(portSource) || 5432;
const user = env.MESH_POSTGRES_USER ?? url?.username ?? "postgres";
const password = env.MESH_POSTGRES_PASSWORD ?? readSecretFile(env.MESH_PROVISION_PASSWORD_FILE);
if (!host || !password) {
@@ -79,9 +88,11 @@ export class PostgresClient {
async createDatabaseAndRole(database: string, role: string, password: string): Promise<void> {
const roles = await this.query("SELECT 1 FROM pg_roles WHERE rolname = " + literal(role));
if (roles.rows.length === 0) {
await this.query(`CREATE ROLE ${ident(role)} WITH LOGIN PASSWORD ${literal(password)}`);
await this.query(`CREATE ROLE ${ident(role)} WITH LOGIN PASSWORD ${literal(password)} VALID UNTIL 'infinity'`);
} else {
await this.query(`ALTER ROLE ${ident(role)} WITH LOGIN PASSWORD ${literal(password)}`);
// VALID UNTIL 'infinity': a password that expired is refused like a wrong one, so the check the
// provisioner runs would report it lost, and only clearing the expiry makes applying it again work.
await this.query(`ALTER ROLE ${ident(role)} WITH LOGIN PASSWORD ${literal(password)} VALID UNTIL 'infinity'`);
}
const dbs = await this.query("SELECT 1 FROM pg_database WHERE datname = " + literal(database));
if (dbs.rows.length === 0) {
@@ -90,6 +101,30 @@ export class PostgresClient {
await this.query(`GRANT ALL PRIVILEGES ON DATABASE ${ident(database)} TO ${ident(role)}`);
}
/**
* Whether `role` can log in to `database` with exactly `password`: the consumer's own view of its
* credential, checked by connecting as it. Read-only. `false` only when the server says so (the
* role, the password or the database is wrong or gone); an unreachable server rejects instead,
* because being unable to ask is not evidence of loss (novox/hq issue 120).
*/
async canConnectAs(database: string, role: string, password: string): Promise<boolean> {
try {
await run(
"psql",
["-h", this.conn.host, "-p", String(this.conn.port), "-U", role, "-d", database,
"-v", "ON_ERROR_STOP=1", "--no-psqlrc", "-tAc", "SELECT 1"],
{ env: { ...process.env, PGPASSWORD: password, PGCONNECT_TIMEOUT: "10" }, timeout: 20_000 },
);
return true;
} catch (err) {
const text = `${(err as { stderr?: string }).stderr ?? ""}`;
if (/password authentication failed|role ".*" does not exist|database ".*" does not exist|not permitted to log in|permission denied for database/i.test(text)) {
return false;
}
throw err;
}
}
/** Drop a database and its owning role, idempotently, after evicting live connections. */
async dropDatabaseAndRole(database: string, role: string): Promise<void> {
await this.query(
+9 -1
View File
@@ -69,6 +69,13 @@
"path": "/var/lib/postgres/grants",
"mode": "0700"
},
{
"id": "store-data",
"type": "directory",
"path": "/var/lib/mesh-store",
"mode": "0700",
"owner": "999:70"
},
{
"id": "server",
"type": "container",
@@ -82,7 +89,7 @@
"5432:5432"
],
"volumes": [
"mesh-store-data:/var/lib/postgresql/data",
"/var/lib/mesh-store:/var/lib/postgresql/data",
"/var/lib/postgres/superuser.secret:/run/secrets/superuser:ro"
]
},
@@ -98,6 +105,7 @@
],
"env": {
"MESH_PROVISION_POSTGRES": "postgres://postgres@127.0.0.1:5432/postgres?sslmode=disable",
"MESH_PROVISION_POSTGRES_PORT": "${seat:mesh-store:5432}",
"MESH_PROVISION_PASSWORD_FILE": "/run/secrets/superuser",
"MESH_BROKER_FILE": "/run/secrets/broker",
"MESH_RECEIVES": "/var/lib/postgres/grants/mesh.json"
+1 -1
View File
@@ -5,7 +5,7 @@
"type": "module",
"private": true,
"dependencies": {
"@novox/mesh-sdk": "^0.1.0"
"@novox/mesh-sdk": "^0.1.1"
},
"devDependencies": {
"@types/node": "^22.0.0",
+5
View File
@@ -45,4 +45,9 @@ runProvisioner("postgres-database", {
await postgres.dropDatabaseAndRole(p.as, p.as);
await announce("module.postgres.database.deprovisioned", { database: p.as });
},
// Asked every minute by the harness: whether the backend still holds this consumer exactly as
// the mesh gave it, so a login lost behind the provisioner's back is made again (novox/hq issue 120).
async holds(p: Provision): Promise<boolean> {
return postgres.canConnectAs(p.as, p.as, p.password);
},
});
+1 -1
View File
@@ -13,7 +13,7 @@
"at": "acme-v02.api.letsencrypt.org",
"port": 443,
"path": "/directory",
"root": ""
"roots": ""
}
}
}
+22 -1
View File
@@ -8,7 +8,7 @@
// order requests were sent, which is what the queue below relies on.
import { createConnection, type Socket } from "node:net";
import { randomBytes } from "node:crypto";
import { createHash, randomBytes } from "node:crypto";
import { readFileSync } from "node:fs";
/** A parsed RESP value. Errors are surfaced as rejected commands, not as this type. */
@@ -113,6 +113,27 @@ export class RedisClient {
await this.command("ACL", "DELUSER", username);
}
/**
* Whether an ACL user exists, is enabled, and accepts exactly this password. Read-only: it asks
* `ACL GETUSER`, which answers nil for an unknown user and otherwise a flat list of fields, among
* them `flags` and `passwords`, the latter as SHA-256 hex. This server keeps no ACL file, so its
* users live in memory and a restart forgets them. This is how the provisioner notices
* (novox/hq issue 120).
*/
async holdsAclUser(username: string, password: string): Promise<boolean> {
const reply = await this.command("ACL", "GETUSER", username);
if (!Array.isArray(reply)) return false;
const field = (name: string): RespValue | undefined => {
const i = reply.indexOf(name);
return i >= 0 ? reply[i + 1] : undefined;
};
const flags = field("flags");
const passwords = field("passwords");
if (!Array.isArray(flags) || !flags.includes("on")) return false;
if (!Array.isArray(passwords)) return false;
return passwords.includes(createHash("sha256").update(password).digest("hex"));
}
close(): void {
if (this.socket) {
this.socket.destroy();
+1 -1
View File
@@ -5,7 +5,7 @@
"type": "module",
"private": true,
"dependencies": {
"@novox/mesh-sdk": "^0.1.0"
"@novox/mesh-sdk": "^0.1.1"
},
"devDependencies": {
"@types/node": "^22.0.0",
+7
View File
@@ -43,4 +43,11 @@ runProvisioner("redis-cache", {
await redis.deleteAclUser(p.as);
await announce("module.redis.cache.deprovisioned", { username: p.as });
},
// This server keeps its ACL users in memory only, so a restart of it forgets every consumer while
// this provisioner keeps running. Asked every minute, so a forgotten user is made again instead
// of every consumer failing to authenticate in silence (novox/hq issue 120).
async holds(p: Provision): Promise<boolean> {
return redis.holdsAclUser(p.as, p.password);
},
});
+16 -6
View File
@@ -2,12 +2,22 @@
"module": "resolv-conf",
"version": "1",
"slug": "resolv",
"requires": ["wildcard-resolution"],
"claims": [{"name": "the-resolver-configuration", "scope": "node"}],
"requires": [
"wildcard-resolution"
],
"claims": [
{
"name": "node-resolver-config",
"scope": "node"
}
],
"resources": [
{"id": "resolv", "type": "file", "path": "/etc/resolv.conf", "mode": "0644",
"content": "# Managed by the mesh.\n#\n# For a machine where nothing else owns this file. On one where systemd-resolved\n# or NetworkManager does, assign that module instead — this one and those claim\n# the same thing, so the mesh refuses the pair rather than letting them take\n# turns overwriting each other, which is the failure this claim exists to stop.\n#\n# The mesh's resolver, and only it — the one line the predecessor wrote on every\n# machine it set up. It answers the mesh's names itself and forwards everything\n# else to upstreams named in its own configuration, never read from this file.\n# This file used to carry a second nameserver as a placeholder for \"whatever\n# this machine used before\"; that was never a fallback for names the mesh does\n# not know — a resolver's second line is asked only when the first does not\n# answer at all — and now that the first answers everything it would be a line\n# nothing ever reached.\nnameserver 127.0.0.1\noptions edns0\n"}
{
"id": "resolv",
"type": "file",
"path": "/etc/resolv.conf",
"mode": "0644",
"content": "# Managed by the mesh.\n#\n# For a machine where nothing else owns this file. On one where systemd-resolved\n# or NetworkManager does, assign that module instead \u2014 this one and those claim\n# the same thing, so the mesh refuses the pair rather than letting them take\n# turns overwriting each other, which is the failure this claim exists to stop.\n#\n# The mesh's resolver, and only it \u2014 the one line the predecessor wrote on every\n# machine it set up. It answers the mesh's names itself and forwards everything\n# else to upstreams named in its own configuration, never read from this file.\n# This file used to carry a second nameserver as a placeholder for \"whatever\n# this machine used before\"; that was never a fallback for names the mesh does\n# not know \u2014 a resolver's second line is asked only when the first does not\n# answer at all \u2014 and now that the first answers everything it would be a line\n# nothing ever reached.\nnameserver 127.0.0.1\noptions edns0\n"
}
]
}
+32 -12
View File
@@ -2,18 +2,38 @@
"module": "resolved-split-dns",
"version": "1",
"slug": "splitdns",
"requires": ["wildcard-resolution"],
"claims": [{"name": "the-resolver-configuration", "scope": "node"}],
"requires": [
"wildcard-resolution"
],
"claims": [
{
"name": "node-resolver-config",
"scope": "node"
}
],
"resources": [
{"id": "drop-in", "type": "directory", "path": "/etc/systemd/resolved.conf.d", "mode": "0755"},
{"id": "route", "type": "file",
"path": "/etc/systemd/resolved.conf.d/mesh.conf", "mode": "0644",
"content": "# Managed by the mesh.\n#\n# **Only the mesh's names.** The tilde makes this a routing domain rather than a\n# search domain: queries under it go to the resolver below, and everything else\n# keeps going wherever this machine already sent it. A resolver that took over\n# all of DNS would be this module claiming the machine's whole network, which\n# is not what it says it claims. The mesh's resolver can forward the rest too;\n# this module is for a machine that wants systemd-resolved to stay in charge of\n# that, and only lends it the mesh's suffix.\n#\n# 127.0.0.1 is where the mesh's resolver answers on every machine — a fixed\n# address, so this file needs to know nothing about this particular machine.\n# systemd-resolved holds .53 and .54 itself, which is why the resolver is on\n# neither, and why the two coexist here.\n[Resolve]\nDNS=127.0.0.1\nDomains=~internal\n"},
{"id": "resolved", "type": "service", "unit": "systemd-resolved.service",
"state": "running", "boot": "enabled", "restart-on": ["route"]}
{
"id": "drop-in",
"type": "directory",
"path": "/etc/systemd/resolved.conf.d",
"mode": "0755"
},
{
"id": "route",
"type": "file",
"path": "/etc/systemd/resolved.conf.d/mesh.conf",
"mode": "0644",
"content": "# Managed by the mesh.\n#\n# **Only the mesh's names.** The tilde makes this a routing domain rather than a\n# search domain: queries under it go to the resolver below, and everything else\n# keeps going wherever this machine already sent it. A resolver that took over\n# all of DNS would be this module claiming the machine's whole network, which\n# is not what it says it claims. The mesh's resolver can forward the rest too;\n# this module is for a machine that wants systemd-resolved to stay in charge of\n# that, and only lends it the mesh's suffix.\n#\n# 127.0.0.1 is where the mesh's resolver answers on every machine \u2014 a fixed\n# address, so this file needs to know nothing about this particular machine.\n# systemd-resolved holds .53 and .54 itself, which is why the resolver is on\n# neither, and why the two coexist here.\n[Resolve]\nDNS=127.0.0.1\nDomains=~internal\n"
},
{
"id": "resolved",
"type": "service",
"unit": "systemd-resolved.service",
"state": "running",
"boot": "enabled",
"restart-on": [
"route"
]
}
]
}
+12
View File
@@ -159,3 +159,15 @@ cd modules/route-adapter && npm test
They hold it to what ADR 0104 says holds it: one file per contribution, a file removed when its
contribution goes, every file it did not write left alone — and the two facts a route file has to
get right, the port the contributor publishes and the address of the machine it is on.
## A body limit
A contribution may say `max-request-body`, in bytes, and the adapter writes it as the predecessor's
own `buffering` middleware, named after the router so the two halves cannot drift. A route that says
nothing gets no middleware and the predecessor's default stands.
This is the one thing the file shape *can* say that a policy cannot, which is why it is written
rather than skipped: the predecessor already served its own registry name this way. A limit that is
not a whole positive number of bytes takes the route with it — written without the limit, the
predecessor would carry exactly what the module said not to carry, and this module would report
success doing it.
+69 -1
View File
@@ -75,6 +75,15 @@ export interface Route {
from: string;
/** Where the predecessor's proxy is to send it. */
target: string;
/**
* The largest request body, in bytes, the predecessor may carry to it — the contribution's
* `max-request-body`. Absent is whatever the predecessor does by default.
*
* Unlike a policy, this file shape *can* say it: the predecessor has a buffering middleware, and
* its own registry route used exactly this. A registry takes image layers in single requests of
* gigabytes, so a route that could not say it would be a name nothing could be pushed to.
*/
maxRequestBody?: number;
}
/** What one pass changed. */
@@ -151,6 +160,24 @@ export function routesFrom(document: unknown, machine: string): { routes: Route[
skipped.push(`${from} asked for ${JSON.stringify(name)}, which is not a name this can write`);
continue;
}
// What this adapter's one file shape cannot say, it skips aloud rather than approximating:
// a backend over its own TLS (the file would send plain http into a TLS listener), a
// path-scoped or refusing or redirecting rule (the file routes whole hosts). The mesh's own
// proxy serves all of these the day it takes over; until then the predecessor's hand-authored
// files keep covering them, exactly as they do today.
const scheme = typeof entry.values?.["scheme"] === "string" ? (entry.values["scheme"] as string).trim().toLowerCase() : "";
if (scheme !== "" && scheme !== "http") {
skipped.push(`${from} asked for route ${name} over ${scheme}, which this file shape cannot say`);
continue;
}
if (typeof entry.values?.["path"] === "string" && (entry.values["path"] as string).trim() !== "") {
skipped.push(`${from} asked for route ${name} scoped to a path, which this file shape cannot say`);
continue;
}
if (entry.values?.["deny"] === true || typeof entry.values?.["redirect"] === "string") {
skipped.push(`${from} asked for route ${name} with a policy this file shape cannot say`);
continue;
}
const port = asPort(entry.values?.["port"]);
if (port === undefined) {
skipped.push(`${from} asked for route ${name} and gave no usable port`);
@@ -158,12 +185,40 @@ export function routesFrom(document: unknown, machine: string): { routes: Route[
}
// Where the mesh says that machine is. Empty means this one, and this one is reached from
// inside the predecessor's container by the machine's own name, not by loopback.
// A limit it cannot honour is a route it does not write — skipped and named, like a port that
// is not one. Written without the limit instead, the predecessor would carry exactly what the
// module said not to carry, and this adapter would report success.
const askedLimit = entry.values?.["max-request-body"];
const limit = asBodyLimit(askedLimit);
if (limit === null) {
skipped.push(
`${from} asked for route ${name} with a max-request-body of ${JSON.stringify(askedLimit)}, ` +
`which is not a whole positive number of bytes`,
);
continue;
}
const at = typeof entry.at === "string" && entry.at.trim() !== "" ? entry.at.trim() : machine;
routes.push({ name, from, target: `http://${at}:${port}` });
routes.push({ name, from, target: `http://${at}:${port}`, ...(limit === undefined ? {} : { maxRequestBody: limit }) });
}
return { routes, skipped };
}
/**
* The body limit a contribution asked for: a number, `undefined` for silence, `null` for unusable.
*
* Three answers rather than two, because "said nothing" and "said something wrong" must not become
* the same route.
*/
function asBodyLimit(value: unknown): number | undefined | null {
if (value === undefined) {
return undefined;
}
if (typeof value !== "number" || !Number.isInteger(value) || value < 1) {
return null;
}
return value;
}
/** The file one route is written to. The prefix is how the mesh recognises its own. */
export function fileNameFor(name: string): string {
return `mesh-${name}.yml`;
@@ -183,6 +238,10 @@ export function routerNameFor(name: string): string {
*/
export function routeFile(route: Route, settings: Settings): string {
const id = routerNameFor(route.name);
// The body limit is a middleware in the predecessor's vocabulary — its `buffering`, with the one
// field the predecessor's own registry route set — named after the router so the two halves cannot
// drift, and written only when the contribution asked for it.
const limited = route.maxRequestBody !== undefined;
return [
marker,
`# ${route.from} contributed this route. It is removed when that contribution goes.`,
@@ -192,10 +251,19 @@ export function routeFile(route: Route, settings: Settings): string {
` entryPoints: [${settings.entrypoint}]`,
` rule: Host(\`${route.name}\`)`,
` service: ${id}`,
...(limited ? [` middlewares: [${id}-body]`] : []),
" tls:",
` certResolver: ${settings.resolver}`,
" domains:",
` - main: ${route.name}`,
...(limited
? [
" middlewares:",
` ${id}-body:`,
" buffering:",
` maxRequestBodyBytes: ${route.maxRequestBody}`,
]
: []),
" services:",
` ${id}:`,
" loadBalancer:",
+63 -2
View File
@@ -22,7 +22,9 @@ async function predecessor(already: Record<string, string> = {}): Promise<Settin
}
/** The contributions file the mesh writes, in the shape the mesh's own proxy also reads. */
function contributed(...given: { from: string; node?: string; at?: string; name: string; port: number }[]) {
function contributed(
...given: { from: string; node?: string; at?: string; name: string; port: number; limit?: unknown }[]
) {
return {
contributions: 1,
requirement: "route",
@@ -30,7 +32,7 @@ function contributed(...given: { from: string; node?: string; at?: string; name:
from: g.from,
node: g.node ?? "control-node",
at: g.at ?? "",
values: { name: g.name, port: g.port },
values: { name: g.name, port: g.port, ...(g.limit === undefined ? {} : { "max-request-body": g.limit }) },
})),
};
}
@@ -205,9 +207,68 @@ test("a contribution it cannot act on is skipped and named", async () => {
assert.deepEqual(routesFrom(undefined, machine).routes, []);
});
// What the file shape cannot say is skipped aloud, never approximated: plain http into a TLS
// listener, a whole-host file for a path-scoped rule, a proxying file for a refusal or redirect.
// The mesh's own proxy serves all of these the day it takes over; until then the predecessor's
// hand-authored files keep covering them.
test("a contribution the file shape cannot say is skipped and says which part", async () => {
const machine = defaults.machine;
const { routes, skipped } = routesFrom({ given: [
{ from: "mailu", values: { name: "mail.example", port: 7443, scheme: "https", insecure: true } },
{ from: "mailu", values: { name: "mail.example", port: 7080, path: "/.well-known/acme-challenge" } },
{ from: "gitea", values: { name: "git.example", path: "/api/internal", deny: true, priority: 100000 } },
{ from: "site", values: { name: "www.example", redirect: "https://example" } },
{ from: "mailu", values: { name: "autoconfig.example", port: 4243 } },
] }, machine);
assert.deepEqual(routes.map((r) => r.name), ["autoconfig.example"]);
assert.equal(skipped.length, 4);
assert.match(skipped[0]!, /over https/);
assert.match(skipped[1]!, /scoped to a path/);
assert.match(skipped[2]!, /scoped to a path/);
assert.match(skipped[3]!, /policy/);
});
// The directory is the predecessor's and the mesh only mounts it. Absent, there is nothing to write
// into — and writing anyway would put route files somewhere nothing reads, reporting success.
test("it refuses when the predecessor's directory is not there, and says why", async () => {
const settings = { ...defaults, dynamic: join(await mkdtemp(join(tmpdir(), "route-adapter-")), "absent") };
await assert.rejects(reconcile([], settings), /is not there.*`dynamic` setting.*mounts it/s);
});
// **A registry is why a route needs to say this.** Image layers arrive as single requests of
// gigabytes, and the predecessor served its own registry name with a `buffering` middleware for
// exactly that reason. The contribution carries the limit as `max-request-body`, the adapter writes
// the middleware the predecessor already understands, named after the router so the two halves
// cannot drift — and writes nothing of the kind for a route that did not ask.
test("a body limit is written as the predecessor's buffering middleware", async () => {
const settings = await predecessor();
const changed = await pass(settings, contributed(
{ from: "registry", name: "images.example", port: 5001, limit: 21474836480 },
{ from: "forge", name: "git.example", port: 2999 },
));
assert.deepEqual(changed.written, ["mesh-git.example.yml", "mesh-images.example.yml"]);
const written = await readFile(join(settings.dynamic, "mesh-images.example.yml"), "utf8");
assert.match(written, /^ {6}middlewares: \[mesh-images-example-body\]$/m);
assert.match(written, /^ {2}middlewares:\n {4}mesh-images-example-body:\n {6}buffering:\n {8}maxRequestBodyBytes: 21474836480$/m);
// The route that asked for nothing carries no middleware — the predecessor's default stands.
const plain = await readFile(join(settings.dynamic, "mesh-git.example.yml"), "utf8");
assert.doesNotMatch(plain, /middlewares|buffering/);
});
// A limit it cannot honour is a route it does not write. Written without it, the predecessor would
// carry exactly what the module said not to carry, and this module would report success.
test("a body limit that is not a whole number of bytes is skipped and named", () => {
for (const limit of ["20g", 0, -1, 1.5, true, null]) {
const { routes, skipped } = routesFrom(
contributed({ from: "registry", name: "images.example", port: 5001, limit }), defaults.machine);
assert.deepEqual(routes, [], `a limit of ${JSON.stringify(limit)} was served`);
assert.equal(skipped.length, 1);
assert.match(skipped[0]!, /max-request-body/);
}
// And a limit the mesh's own proxy would accept is carried through, as a number.
const { routes } = routesFrom(
contributed({ from: "registry", name: "images.example", port: 5001, limit: 1024 }), defaults.machine);
assert.equal(routes[0]?.maxRequestBody, 1024);
});
+48 -9
View File
@@ -18,10 +18,12 @@
"route": "/var/lib/route-proxy/routes/mesh.json"
},
"requires": [
"acme-ca"
"acme-ca",
"internal-acme-ca"
],
"binds": {
"acme-ca": "/var/lib/route-proxy/acme-ca.json"
"acme-ca": "/var/lib/route-proxy/acme-ca.json",
"internal-acme-ca": "/var/lib/route-proxy/internal-acme-ca.json"
},
"listens": [
{
@@ -67,7 +69,14 @@
"type": "file",
"path": "/var/lib/route-proxy/acme.env",
"mode": "0600",
"content": "ACME_DIRECTORY=https://${bound:acme-ca:at}:${bound:acme-ca:port}${bound:acme-ca:path}\nACME_ROOTS=https://${bound:acme-ca:at}:${bound:acme-ca:port}${bound:acme-ca:roots}\n"
"content": "ACME_DIRECTORY=https://${bound:acme-ca:at}:${bound:acme-ca:port}${bound:acme-ca:path}\nACME_ROOTS=https://${bound:acme-ca:at}:${bound:acme-ca:port}${bound:acme-ca:roots}\nACME_ROOTS_PATH=${bound:acme-ca:roots}\n"
},
{
"id": "internal-acme-env",
"type": "file",
"path": "/var/lib/route-proxy/internal-acme.env",
"mode": "0600",
"content": "INTERNAL_ACME_DIRECTORY=https://${bound:internal-acme-ca:at}:${bound:internal-acme-ca:port}${bound:internal-acme-ca:path}\nINTERNAL_ACME_ROOTS=https://${bound:internal-acme-ca:at}:${bound:internal-acme-ca:port}${bound:internal-acme-ca:roots}\nINTERNAL_ACME_ROOTS_PATH=${bound:internal-acme-ca:roots}\n"
},
{
"id": "trust",
@@ -85,20 +94,43 @@
"args": [
"sh",
"-c",
"for i in $(seq 1 60); do wget -q -T 10 --no-check-certificate -O /ca/root.crt \"$ACME_ROOTS\" && grep -q 'BEGIN CERTIFICATE' /ca/root.crt && exit 0; sleep 2; done; echo \"the authority at $ACME_ROOTS did not serve its roots within two minutes\" >&2; exit 1"
"if [ -z \"$ACME_ROOTS_PATH\" ]; then cp /etc/ssl/certs/ca-certificates.crt /ca/root.crt; exit 0; fi; for i in $(seq 1 60); do wget -q -T 10 --no-check-certificate -O /ca/root.crt \"$ACME_ROOTS\" && grep -q 'BEGIN CERTIFICATE' /ca/root.crt && exit 0; sleep 2; done; echo \"the authority at $ACME_ROOTS did not serve its roots within two minutes\" >&2; exit 1"
],
"restart-on": [
"acme-env"
]
},
{
"id": "internal-trust",
"type": "container",
"name": "route-proxy-internal-trust",
"artifact": "trust",
"run-once": true,
"network": "host",
"env-file": [
"/var/lib/route-proxy/internal-acme.env"
],
"volumes": [
"/var/lib/route-proxy/ca:/ca"
],
"args": [
"sh",
"-c",
"if [ -z \"$INTERNAL_ACME_ROOTS_PATH\" ]; then cp /etc/ssl/certs/ca-certificates.crt /ca/internal-root.crt; exit 0; fi; for i in $(seq 1 60); do wget -q -T 10 --no-check-certificate -O /ca/internal-root.crt \"$INTERNAL_ACME_ROOTS\" && grep -q 'BEGIN CERTIFICATE' /ca/internal-root.crt && exit 0; sleep 2; done; echo \"the authority at $INTERNAL_ACME_ROOTS did not serve its roots within two minutes\" >&2; exit 1"
],
"restart-on": [
"internal-acme-env"
]
},
{
"id": "server",
"type": "container",
"name": "route-proxy",
"image": "mesh-route-proxy@sha256:0000000000000000000000000000000000000000000000000000000000000000",
"artifact": "server",
"network": "host",
"env-file": [
"/var/lib/route-proxy/acme.env"
"/var/lib/route-proxy/acme.env",
"/var/lib/route-proxy/internal-acme.env"
],
"volumes": [
"/var/lib/route-proxy/routes:/routes:ro",
@@ -110,11 +142,14 @@
"LISTEN": ":80",
"TLS_LISTEN": ":443",
"ACME_CACHE": "/acme",
"ACME_CA_BUNDLE": "/ca/root.crt"
"ACME_CA_BUNDLE": "/ca/root.crt",
"INTERNAL_ACME_CA_BUNDLE": "/ca/internal-root.crt"
},
"restart-on": [
"trust",
"acme-env"
"acme-env",
"internal-trust",
"internal-acme-env"
]
}
],
@@ -123,7 +158,11 @@
{
"name": "server",
"kind": "image",
"from": "Dockerfile"
"from": "Dockerfile",
"context": {
"repository": "https://git.novox.be/novox/mesh-controller.git",
"ref": "main"
}
},
{
"name": "trust",
+7 -1
View File
@@ -29,6 +29,12 @@
"path": "/var/lib/searxng-module",
"mode": "0700"
},
{
"id": "valkey-data",
"type": "directory",
"path": "/var/lib/searxng-module/valkey-data",
"mode": "0700"
},
{
"id": "server-env",
"type": "file",
@@ -56,7 +62,7 @@
"warning"
],
"volumes": [
"searxng-valkey-data:/data"
"/var/lib/searxng-module/valkey-data:/data"
]
},
{
+182 -59
View File
@@ -2,72 +2,195 @@
"module": "showcase",
"version": "1",
"slug": "show",
"capabilities": ["container-runtime"],
"provides": [{ "name": "greeting", "scope": "mesh" }],
"serves": { "greeting": { "path": "/greeting" } },
"requires": ["postgres-database"],
"binds": { "postgres-database": "/var/lib/showcase/database.json" },
"secrets": { "postgres-database": "/var/lib/showcase/database.secret" },
"own-secrets": { "broker": "/var/lib/mesh/showcase/broker" },
"claims": [{ "name": "the-showcase", "scope": "node" }],
"emits": ["module.showcase.acknowledged"],
"consumes": ["module.showcase.greeted"],
"capabilities": [
"container-runtime"
],
"provides": [
{
"name": "greeting",
"scope": "mesh"
}
],
"serves": {
"greeting": {
"path": "/greeting"
}
},
"requires": [
"postgres-database"
],
"binds": {
"postgres-database": "/var/lib/showcase/database.json"
},
"secrets": {
"postgres-database": "/var/lib/showcase/database.secret"
},
"own-secrets": {
"broker": "/var/lib/mesh/showcase/broker"
},
"claims": [
{
"name": "the-showcase",
"scope": "node"
}
],
"emits": [
"module.showcase.acknowledged"
],
"consumes": [
"module.showcase.greeted"
],
"listens": [
{ "port": 8080, "protocol": "tcp", "from": "mesh",
"why": "the port the daemon itself listens on. The mesh assigns the machine-side number and tells consumers that one (ADR 0038)" }
{
"port": 8080,
"protocol": "tcp",
"from": "mesh",
"why": "the port the daemon itself listens on. The mesh assigns the machine-side number and tells consumers that one (ADR 0038)"
}
],
"build": {
"artifacts": [
{ "name": "code", "kind": "bundle", "language": "typescript",
"entrypoints": ["index.js", "tools/index.js", "provisioner/index.js",
"daemon/index.js", "step/index.js", "report/index.js"] },
{ "name": "files", "kind": "archive", "from": "files" },
{ "name": "helper", "kind": "upstream",
"from": "alpine@sha256:28bd5fe8b56d1bd048e5babf5b10710ebe0bae67db86916198a6eec434943f8b" }
{
"name": "code",
"kind": "bundle",
"language": "typescript",
"entrypoints": [
"index.js",
"tools/index.js",
"provisioner/index.js",
"daemon/index.js",
"step/index.js",
"report/index.js"
]
},
{
"name": "files",
"kind": "archive",
"from": "files"
},
{
"name": "helper",
"kind": "upstream",
"from": "alpine@sha256:28bd5fe8b56d1bd048e5babf5b10710ebe0bae67db86916198a6eec434943f8b"
}
]
},
"resources": [
{ "id": "account", "type": "user", "name": "showcase", "shell": "/usr/bin/nologin",
"home": "/var/lib/showcase" },
{ "id": "logs", "type": "access", "path": "/var/log", "mode": "0755" },
{ "id": "mesh-state", "type": "directory", "path": "/var/lib/mesh/showcase", "mode": "0700" },
{ "id": "state", "type": "directory", "path": "/var/lib/showcase", "mode": "0755" },
{ "id": "settings", "type": "file", "path": "/var/lib/showcase/showcase.env", "mode": "0600",
"content": "SHOWCASE_GREETING=hello\nSHOWCASE_EVERY_SECONDS=30\nSHOWCASE_STATE=/var/lib/showcase\nSHOWCASE_DATABASE=${bound:postgres-database:at}\nSHOWCASE_LISTEN=${port:8080}\n" },
{ "id": "packed", "type": "archive", "path": "/opt/showcase", "artifact": "files" },
{ "id": "net", "type": "network", "name": "showcase" },
{ "id": "tooling", "type": "package", "package": "jq" },
{ "id": "migrate", "type": "process", "name": "showcase-migrate", "artifact": "code",
"run": ["node", "step/index.js"], "run-once": true,
"env-file": ["/var/lib/showcase/showcase.env"] },
{ "id": "server", "type": "process", "name": "showcase", "artifact": "code",
"run": ["node", "daemon/index.js"], "user": "showcase",
"env-file": ["/var/lib/showcase/showcase.env"],
"restart-on": ["settings"] },
{ "id": "reporting", "type": "process", "name": "showcase-report", "artifact": "code",
"run": ["node", "report/index.js"], "schedule": "0 3 * * *",
"env-file": ["/var/lib/showcase/showcase.env"] },
{ "id": "tools", "type": "container", "name": "mesh-showcase", "artifact": "helper",
{
"id": "account",
"type": "user",
"name": "showcase",
"shell": "/usr/bin/nologin",
"home": "/var/lib/showcase"
},
{
"id": "logs",
"type": "access",
"path": "/var/log",
"mode": "0755"
},
{
"id": "mesh-state",
"type": "directory",
"path": "/var/lib/mesh/showcase",
"mode": "0700"
},
{
"id": "state",
"type": "directory",
"path": "/var/lib/showcase",
"mode": "0755"
},
{
"id": "settings",
"type": "file",
"path": "/var/lib/showcase/showcase.env",
"mode": "0600",
"content": "SHOWCASE_GREETING=hello\nSHOWCASE_EVERY_SECONDS=30\nSHOWCASE_STATE=/var/lib/showcase\nSHOWCASE_DATABASE=${bound:postgres-database:at}\nSHOWCASE_LISTEN=${port:8080}\n"
},
{
"id": "packed",
"type": "archive",
"path": "/opt/showcase",
"artifact": "files"
},
{
"id": "net",
"type": "network",
"name": "showcase"
},
{
"id": "tooling",
"type": "package",
"package": "jq"
},
{
"id": "migrate",
"type": "process",
"name": "showcase-migrate",
"artifact": "code",
"run": [
"node",
"step/index.js"
],
"run-once": true,
"env-file": [
"/var/lib/showcase/showcase.env"
]
},
{
"id": "server",
"type": "process",
"name": "showcase",
"artifact": "code",
"run": [
"node",
"daemon/index.js"
],
"user": "showcase",
"env-file": [
"/var/lib/showcase/showcase.env"
],
"restart-on": [
"settings"
]
},
{
"id": "reporting",
"type": "process",
"name": "showcase-report",
"artifact": "code",
"run": [
"node",
"report/index.js"
],
"schedule": "0 3 * * *",
"env-file": [
"/var/lib/showcase/showcase.env"
]
},
{
"id": "tools",
"type": "container",
"name": "mesh-showcase",
"artifact": "helper",
"network": "showcase",
"volumes": ["/var/lib/mesh/showcase/broker:/run/secrets/broker:ro"],
"env": { "MESH_BROKER_FILE": "/run/secrets/broker" },
"args": ["sleep", "infinity"] }
"volumes": [
"/var/lib/mesh/showcase/broker:/run/secrets/broker:ro"
],
"env": {
"MESH_BROKER_FILE": "/run/secrets/broker"
},
"args": [
"sleep",
"infinity"
]
}
],
"seats": [
{
"name": "the-showcase",
"scope": "node"
}
]
}
+26
View File
@@ -0,0 +1,26 @@
{
"module": "ssh-client",
"version": "1",
"resources": [
{
"id": "openssh",
"type": "package",
"package": "openssh"
},
{
"id": "ssh-dir",
"type": "directory",
"path": "${machine:account-home}/.ssh",
"owner": "${machine:account}",
"mode": "0700"
}
],
"facts": {
"ssh-config": {
"path": ".ssh/config",
"home": true,
"shared": true,
"template": "# The mesh's Host blocks — every other node, so `ssh <node>` reaches it as the\n# right account. This region is replaced whenever a node joins, leaves or is\n# renamed; the rest of this file is yours and is kept untouched.\n{{range .Machines}}{{if ne .Name $.Node}}\nHost {{.Name}} {{.FQDN}}\n HostName {{.FQDN}}\n{{if .Account}} User {{.Account}}\n{{end}}{{end}}{{end}}"
}
}
}
+39
View File
@@ -0,0 +1,39 @@
{
"module": "sshd",
"version": "1",
"capabilities": [
"package-manager",
"service-manager"
],
"listens": [
{
"port": 22,
"protocol": "tcp",
"from": "anywhere",
"why": "the operator's own door. From anywhere because the machines that need it are exactly the ones not on the mesh yet \u2014 and locking the operator out is the one failure a firewall must never arrange"
}
],
"resources": [
{
"id": "package",
"type": "package",
"package": "openssh"
},
{
"id": "config",
"type": "file",
"path": "/etc/ssh/sshd_config.d/10-mesh.conf",
"mode": "0644",
"content": "# Managed by the mesh (module sshd). Replaced on every push; edit the catalogue instead.\nPort 22\nPermitRootLogin no\nPasswordAuthentication no\nPubkeyAuthentication yes\nKbdInteractiveAuthentication no\nUsePAM yes\nX11Forwarding no\nPrintMotd no\nAcceptEnv LANG LC_*\n"
},
{
"id": "run",
"type": "service",
"unit": "sshd.service",
"state": "running",
"restart-on": [
"config"
]
}
]
}
+8
View File
@@ -8,12 +8,20 @@
{
"name": "acme-ca",
"scope": "mesh"
},
{
"name": "internal-acme-ca",
"scope": "mesh"
}
],
"serves": {
"acme-ca": {
"path": "/acme/acme/directory",
"roots": "/roots.pem"
},
"internal-acme-ca": {
"path": "/acme/acme/directory",
"roots": "/roots.pem"
}
},
"listens": [
+36
View File
@@ -0,0 +1,36 @@
{
"module": "systemd-networkd",
"version": "1",
"capabilities": [
"package-manager",
"service-manager"
],
"claims": [
{
"name": "node-uplink",
"scope": "node"
}
],
"resources": [
{
"id": "package",
"type": "package",
"package": "systemd"
},
{
"id": "config",
"type": "file",
"path": "/etc/systemd/network/00-mesh0.network",
"mode": "0644",
"content": "# Managed by the mesh (module systemd-networkd). Replaced on every push; edit\n# the catalogue instead.\n#\n# This machine's uplink is systemd-networkd's, and the mesh asks one thing of it\n# here (novox/hq ADR 0117): leave the private network's interface alone. mesh0\n# is the mesh's; the mesh brings it up and configures it itself. The mesh never\n# declares a link, an address, a route, a wireless network or its credentials,\n# nor a network file for any of this machine's own interfaces \u2014 those are\n# the operator's, and the link they make is the only channel the mesh reaches\n# this machine over.\n#\n# 00-: networkd applies the first .network file, in alphanumeric order across\n# every directory, that matches an interface, and ignores every later one even\n# if it matches too (systemd.network(5), [Match]). A catch-all of the operator's\n# \u2014 Name=*, Type=ether, a file with no [Match] at all \u2014 sorted before\n# this one would claim mesh0 first. 00 sorts before every numbered prefix the\n# man page recommends.\n#\n# Unmanaged=yes: \"no attempts are made to bring up or configure matching links,\n# equivalent to when there are no matching network files\" (systemd.network(5),\n# [Link], since 233). A match that ends the search, and does nothing else.\n#\n# No DNS setting, because none is needed: networkd never writes\n# /etc/resolv.conf. What it learns from a lease it hands only to\n# systemd-resolved, and the resolver file stays whatever resolv-conf wrote.\n# Whether resolved runs, and what it does with that, is the resolver\n# configuration's question, not the uplink's.\n#\n# The service is reloaded when this file changes, never restarted: a restart\n# drops the links networkd holds, this machine's channel to the mesh among them.\n[Match]\nName=mesh0\n\n[Link]\nUnmanaged=yes\n"
},
{
"id": "service",
"type": "service",
"unit": "systemd-networkd.service",
"reload-on": [
"config"
]
}
]
}
-91
View File
@@ -1,91 +0,0 @@
// The Verdaccio (npm registry) client — verdaccio's own code, living in the module (novox/hq
// ADR 0039). Both this module's tools and its events entrypoint import it, and nothing outside
// verdaccio does.
import { readFileSync } from "node:fs";
export interface VerdaccioPackage {
name: string;
version?: string;
description?: string;
time?: string;
}
export interface PackageInfo {
name: string;
latest?: string;
versions: string[];
description?: string;
modified?: string;
}
/** The settings-merged config the mesh delivers (novox/hq ADR 0046): { url, apiKey, token, password, user, ... }. */
function meshConfig(file?: string): Record<string, string> {
if (!file) return {};
try { return JSON.parse(readFileSync(file, "utf8")) as Record<string, string>; }
catch { return {}; }
}
export class VerdaccioClient {
readonly baseUrl: string;
// A bearer token is optional: package listing and reading are public on most registries, so the
// token is sent only when configured, for a registry that gates reads behind auth.
constructor(
url: string,
private readonly token?: string,
) {
this.baseUrl = url.replace(/\/+$/, "");
}
/**
* Build from the module's resolved environment. The URL is MESH_VERDACCIO_URL (or the local
* port); an optional MESH_VERDACCIO_TOKEN authenticates. Throws when no URL is configured.
*/
static fromEnv(env: NodeJS.ProcessEnv = process.env): VerdaccioClient {
const cfg = meshConfig(env.MESH_VERDACCIO_CONFIG_FILE);
const url = cfg.url ?? (env.MESH_VERDACCIO_URL ?? `http://127.0.0.1:${env.VERDACCIO_PORT ?? "4873"}`);
if (!url) throw new Error("no verdaccio URL — set MESH_VERDACCIO_URL");
return new VerdaccioClient(url, cfg.token ?? env.MESH_VERDACCIO_TOKEN);
}
private async getJson<T>(path: string): Promise<T> {
const res = await fetch(`${this.baseUrl}${path}`, {
headers: {
Accept: "application/json",
...(this.token ? { Authorization: `Bearer ${this.token}` } : {}),
},
});
if (!res.ok) throw new Error(`Verdaccio ${path}: ${res.status} ${await res.text()}`);
return res.json() as Promise<T>;
}
/**
* Every package the registry hosts, from Verdaccio's own web API — the same list its UI shows.
* Each entry carries the latest version and the time it was last published.
*/
async listPackages(): Promise<VerdaccioPackage[]> {
const raw = await this.getJson<any[]>("/-/verdaccio/data/packages");
return (raw ?? []).map((p) => ({
name: p.name,
version: p.version ?? p["dist-tags"]?.latest,
description: p.description,
time: p.time?.modified ?? p.time,
}));
}
/**
* The full detail of one package — its dist-tags, every published version, and timestamps —
* from the standard npm packument endpoint (`GET /<name>`).
*/
async getPackageInfo(name: string): Promise<PackageInfo> {
const doc = await this.getJson<any>(`/${encodeURIComponent(name).replace(/%2F/g, "/")}`);
return {
name: doc.name ?? name,
latest: doc["dist-tags"]?.latest,
versions: Object.keys(doc.versions ?? {}),
description: doc.description,
modified: doc.time?.modified,
};
}
}
-45
View File
@@ -1,45 +0,0 @@
// verdaccio's events. The tool runtime imports this once the broker is bound.
//
// Emits (novox/hq ADR 0041/0042):
// module.verdaccio.package.published — a new package version was published to the registry
//
// A genuinely useful signal: a package was just published, so anything on the mesh that pins,
// mirrors or announces dependency releases can react without polling the registry. Verdaccio has
// no publish webhook, so the module discovers it by diffing the package list's latest versions.
//
// The polling is deliberately unhurried: a publish a minute late is still the event, whereas
// hammering the registry for immediacy nobody asked for is not.
import { emit } from "@novox/mesh-sdk/events";
import { VerdaccioClient } from "./client.js";
const verdaccio = VerdaccioClient.fromEnv();
// The latest version we have seen per package name. Primed silently on the first look so a registry
// that was already populated when this started does not announce its whole catalog as freshly
// published.
const latest = new Map<string, string>();
let primed = false;
async function pollPackages(): Promise<void> {
const packages = await verdaccio.listPackages();
for (const pkg of packages) {
if (!pkg.version) continue;
const known = latest.get(pkg.name);
if (known !== pkg.version) {
// A name we have not seen, or a name whose latest version moved — both are a publish.
if (primed) await emit("module.verdaccio.package.published", { name: pkg.name, version: pkg.version });
latest.set(pkg.name, pkg.version);
}
}
primed = true;
}
const tick = (fn: () => Promise<void>, everyMs: number): void => {
const run = (): void => void fn().catch((err) => console.error(`[verdaccio] ${err}`));
setInterval(run, everyMs);
run();
};
tick(pollPackages, 60_000);
console.log("[verdaccio] watching the registry for newly published packages");
-130
View File
@@ -1,130 +0,0 @@
{
"module": "verdaccio",
"version": "1",
"slug": "verdacc",
"capabilities": [
"container-runtime"
],
"emits": [
"module.verdaccio.package.published"
],
"own-secrets": {
"broker": "/var/lib/mesh/verdaccio/broker"
},
"listens": [
{
"port": 4873,
"protocol": "tcp",
"from": "mesh",
"why": "the package registry, for installs and publishes"
}
],
"resources": [
{
"id": "mesh-state",
"type": "directory",
"path": "/var/lib/mesh/verdaccio",
"mode": "0700"
},
{
"id": "conf",
"type": "directory",
"path": "/services/verdaccio/conf",
"mode": "0755",
"owner": "10001:10001"
},
{
"id": "storage",
"type": "directory",
"path": "/services/verdaccio/storage",
"mode": "0700",
"owner": "10001:10001"
},
{
"id": "config",
"type": "file",
"path": "/services/verdaccio/conf/config.yaml",
"mode": "0644",
"content": "storage: /verdaccio/storage\nauth:\n htpasswd:\n file: /verdaccio/conf/htpasswd\n max_users: 10\nuplinks:\n npmjs:\n url: https://registry.npmjs.org/\npackages:\n \"**\":\n access: $all\n publish: $authenticated\n proxy: npmjs\nserver:\n keepAliveTimeout: 60\n maxBodySize: 10mb\nmiddlewares:\n audit:\n enabled: true\nlog:\n type: stdout\n format: pretty\n level: http\n"
},
{
"id": "server",
"type": "container",
"name": "verdaccio",
"image": "verdaccio/verdaccio@sha256:7b067a47ae51fb9dff3dcdce60ec0a2cbd7650c208cb4b9f6d37cb1b09b39d43",
"ports": [
"4873"
],
"volumes": [
"/services/verdaccio/storage:/verdaccio/storage",
"/services/verdaccio/conf:/verdaccio/conf"
]
},
{
"id": "runtime-config",
"type": "file",
"path": "/var/lib/mesh/verdaccio/config.json",
"mode": "0600",
"content": "{}\n",
"merge": "json"
},
{
"id": "runtime",
"type": "container",
"name": "mesh-verdaccio",
"network": "host",
"volumes": [
"/var/lib/mesh/verdaccio/broker:/run/secrets/broker:ro",
"/var/lib/mesh/verdaccio/config.json:/run/config/config.json:ro"
],
"env": {
"MESH_BROKER_FILE": "/run/secrets/broker",
"MESH_VERDACCIO_URL": "http://127.0.0.1:4873",
"MESH_VERDACCIO_CONFIG_FILE": "/run/config/config.json"
},
"restart-on": [
"runtime-config"
],
"artifact": "runtime"
}
],
"requires": [
"route"
],
"contributes": {
"route": {
"label": "npm",
"port": 4873
}
},
"binds": {
"route": "/var/lib/mesh/verdaccio/route.json"
},
"provides": [
{
"name": "package-registry",
"scope": "mesh"
}
],
"build": {
"on": [
{
"arg": "BUILD_BASE",
"module": "mesh-tools",
"artifact": "build"
},
{
"arg": "RUNTIME_BASE",
"module": "mesh-tools",
"artifact": "runtime"
}
],
"artifacts": [
{
"name": "runtime",
"kind": "image",
"from": "Dockerfile"
}
]
}
}
-14
View File
@@ -1,14 +0,0 @@
{
"name": "@novox/module-verdaccio",
"version": "0.1.0",
"description": "verdaccio — private npm registry. Its API client, tools and events live here (novox/hq ADR 0039).",
"type": "module",
"private": true,
"dependencies": {
"@novox/mesh-sdk": "^0.1.0"
},
"devDependencies": {
"@types/node": "^22.0.0",
"typescript": "^5.6.0"
}
}
-35
View File
@@ -1,35 +0,0 @@
// verdaccio's tools — its own code (novox/hq ADR 0039), importing verdaccio's own client. They
// return structured data; the mesh serves them through the sdk's tool harness.
import { registerModuleTools, type ToolDefinition } from "@novox/mesh-sdk/tools";
import { VerdaccioClient } from "../client.js";
export function getVerdaccioTools(verdaccio: VerdaccioClient): ToolDefinition[] {
return [
{
name: "verdaccio_list_packages",
description: "List every package hosted on the private npm registry, with each one's latest version.",
input: {},
run: async () => {
const packages = await verdaccio.listPackages();
return { count: packages.length, packages };
},
},
{
name: "verdaccio_package_info",
description: "Details of one package on the registry: its latest tag, all published versions, and description.",
input: { name: { type: "string", description: "the package name, e.g. '@novox/mesh-sdk'" } },
run: async (args) => verdaccio.getPackageInfo(String(args.name)),
},
];
}
// The tools exist only when a registry URL is configured; otherwise verdaccio contributes none
// rather than failing the whole runtime.
registerModuleTools("verdaccio", (env) => {
try {
return getVerdaccioTools(VerdaccioClient.fromEnv(env));
} catch {
return [];
}
});
-12
View File
@@ -1,12 +0,0 @@
{
"compilerOptions": {
"target": "ES2022",
"module": "NodeNext",
"moduleResolution": "NodeNext",
"strict": true,
"esModuleInterop": true,
"skipLibCheck": true,
"noEmit": true
},
"include": ["client.ts", "index.ts", "tools/index.ts"]
}