One helper serves every part that reads whole states (ADR 0580 clause 3): the assigner's own watched mirror goes, its
compare reads the large states from stdio.WatchedStates and each manifest once per digest. mesh-sdk pinned at 8311ecb.
Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
Issue 1585, S6-S8: 192 judge-only declarations (the strict parser, every manifest problem function, the tables)
moved whole by an AST mover, methods becoming functions; 163 helpers composition keeps too are copied. 245 tests
move with them, reading the captured catalogue through a copy of internal/beside. judge.WithTemplatesFrom and
ReadTreeManifest judge what mesh-sdk's readers now only fill. mesh-delivery's own tests decode. Pinned to mesh-sdk 25a2cec.
Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
Issue 1585, S11: world and wants read with catalogue.Decode; a trunk refusal is held, named and handed to the
composer in its input; a proposal's composition read that fails is said, not dropped; a reader that cannot read a
changed field is a warning, no longer a failure.
Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
The holder writes a node's changed module keys, then its node key as the commit by compare-and-set at
the revision the proposal was laid over, then the placement.<node> mirror for readers not yet delivered.
Every reader assembles through mesh-sdk's PlacementKeys, which reads both forms.
On novox the asks of postgres, keycloak and mongodb were judged stale against the controller's compositions, taken away and made again under a new id every pass (8 s), so no answer could count; and every put is a change the node composes again on.
Quick fix: the stale check keeps a record ask whose digest the node's have waits on; the pass skips a put whose value the bucket holds.
Thorough fix beside it (not here, another branch owns the file): the node-engine's composer composes again only when a want's value changed.
Rule 8 asked a key for every `requires`, and the plans key only a handed module's needs and what a helper answers:
grafana waited on ace for oidc-client, which the controller still makes, and mesh-build-go for go-toolchain on its own
machine, for ever. MissingProvisions now takes catalogue.Awaited, the one rule the plan composition and mesh-provisioner
key by, judged against the providers placed somewhere.
Quick fix: drop the missing-key check. Thorough fix (chosen): the shared rule in mesh-sdk, so the three parts cannot
drift again. Needs mesh-sdk at 30098229b831.
mesh/check checking 8fe6442c4754 on c104e7500d21 | novox/mesh-delivery#118 in group fix/held-module-and-assigner-contract
A pin made by hand, as the condition's Needs asks, changed nothing the assigner looked at, so its urgent condition
never cleared, and a restart left every one raised. The assigner now reads the controller's pins (placement-export,
asked again only after the controller's compositions change, never on a timer) and clears a pin's condition once
they agree with the placement's after being seen differing, as a module's place already did. The first pass takes
over the call conditions a previous process raised, pins and places alike, and retries at the next pass when the
findings cannot be read.
The assigner added "why" to every assign, unassign, pin and unpin since 2026-10-28; the controller's seat contract
takes no such argument, so it refused each one and did nothing. No approved placement reached the controller's
assignments, no bus account was minted for a newly placed module, and the refusal was only a journal line.
Quick fix: drop "why" from ControllerCalls. Thorough fix: a contract test that every argument ControllerCalls gives,
for every verb, is declared (and every required one given) in catalogue.ControllerVerbs, plus an urgent condition per
pair whose call was refused (verb, node, module, the controller's words), cleared when a later call is answered or the
two sources agree. Chosen: both here; a caller-side check of arguments against the seat's contract for every seat
call belongs in mesh-sdk and is noted for the lead. The warrant stays in the assigner's own record (placement-changed,
the hand-act log).
Every module pins mesh-sdk eff81f1220e2 (in review).
mesh/check pass d54eab4052d1 on c1b4533201c8 | unit 3318 run, 1951 kept, 38 skipped | integration 3994 run, 2125 kept, 38 skipped | over bound: unit, …
On novox on 2026-10-29 the assigner wrote no want for de-spiegel and link2pay (rule 5: no provider of route runs
healthy) nor photos (rule 5: no provider of s3-bucket), and none where no node of a module was online (rule 3). A
module with no want is on no list and nothing applies it; a site runs while the proxy in front of it is down.
- Quick fix: skip rules 3 and 5 for the three modules' first wants. Leaves rule 5 holding every move and the
canary holding new assignments.
- Thorough fix, chosen: rule 5 holds no want; it is said per node (Outcome.Unreached, journaled once per words by the
pass). The canary (rule 3) no longer holds the first want of a node that does not run the module. A node that runs
it keeps the canary order: at an older composed and granted fingerprint of the new form its first want stays at what
it runs; at an old-form build it waits for the canary as before (issue 1945); at an older fingerprint with no
composition or grant it gets F. This departs from ADR 0429 §6, which makes every provider healthy a rule of every want.
Tests: rows changed in the existing rule tests, and rows for today's modules (route-proxy down: de-spiegel and
link2pay wanted; minio not healthy: photos moves; novox offline: a first want written, a move kept back).
mesh/check failed: shanks: novox/mesh-delivery#105 does not merge into the trunk: files in conflict: modules/mesh-delivery-assigner/go.mod, modules/me…
Why: from the lease write the grants part's state grants is the source and the controller's bucket otherwise; the assigner follows the lease key (a pass only when the source changes), and neither bucket stops its start. Needs the group's mesh-sdk pin (GrantSource, LeaseHandedAway, the assigner a named reader of the lease).
mesh/check failed: integration: meshbuildgo/cmd/mesh-build-go.TestAComposeAskIsBuiltOnceAndAnsweredOnTheBus: the compose ask mvsg85b34d25b4 never ende…
mesh-grants declares each module's served settings (catalogue.ServedSettings), so
the user list grants every other node's agent those keys of mesh-config_resolved
alone: shanks then settles what novox's mailu serves. The assigner's check of
every node reads no build, archive or served value: it composes as before.
mesh/check being carried to bb63d83f849e on c38e81c7f5e9 | from 2915d74822cd
After the placement seed's revert the rules that hold a want (a hold, rules 3 and 5) kept g14's and shanks's Self wants: the held branch copied them whole, and the rule branch wrote a want naming no composition, which EncodeWant refuses. Module.Self, set once from the placement, is now the one source, and every want the rules write or keep is derived from it: an unplaced node's kept want names the controller's composition of the same fingerprint, asked as any record move; with none, it stays and rule 7 says why.
Review round 2 of 16e4e61: gating the first-pass clear on a failed have left a
condition an earlier process raised standing for good once its node recovered
while no assigner watched. The first pass now clears it for every module and
every node the presence bucket knows, a quiet clear at the router for one that
is not open.
Review of 2ee7589: a node that sleeps after running F healthy kept its stale
have as a witness, so F stayed unheld while it failed on the online nodes; and a
failed-alone condition was never cleared once its node left the module.
A witness must now be online by the watchman's presence. F failing on a second
node is no longer one node's failure: the hold and the set-back stand (ADR 0557
version 6, §2). A remembered failed-alone key whose node is no longer one of the
module's nodes is cleared once; the first pass after a start clears it for every
node the presence bucket knows, for a module with a failed have.
On ace, 2026-10-26 21:19, a newly assigned mesh-build-node failed its first
health check at 8a98386f (later 31be2f24) while shanks, g14 and novox had run it
healthy for an hour. Being first by name, ace was the canary, and the assigner
held the version and moved the three healthy nodes back.
A failed health check of the newest fingerprint now holds it only when no other
node's have names it healthy (ADR 0557 §2 keeps ADR 0429 §6's hold and set-back
for that case). Otherwise the failure is that node's (ADR 0557 §1): nothing is
held or set back, the node keeps its want, and the assigner raises the node's
condition once (failed-alone.<node>.<module>), naming the node, the module, the
fingerprint and the have's words, and clears it once when it ends; the first
pass after a start clears what a previous process may have raised.
mesh/check failed: ace: novox/mesh-catalog#505 does not merge into the trunk: files in conflict: .gitignore | 2e0013ec2500 on df8b9ac3b1b5 | unit 1245…
One commit for the branch feat/1126-nodes-compose of mesh-delivery: its 16 commits squashed so its history does not carry the 17.8 MB compiled mesh-delivery-assigner a commit added by mistake and a later one removed (issue 1577). The tree equals the former head ff6c038d. The operator allowed this rewrite of the feature branch on 2026-10-26.
jschoubben
mesh-tools