A machine says what it is made of, and says it once a day unasked (issue 160) #92

Merged
mesh-admin merged 2 commits from fix/160-a-machine-says-what-it-is into main 2026-10-12 21:07:05 +00:00
Contributor

What

  • internal/profile/machine.go: the profile now carries machine, what the machine is made of: memory, each local filesystem (mount, device, type, size, free), processor model, cores and threads, distribution and version, kernel release, virtual or physical (systemd-detect-virt), boot time (uptime is the time since) and timezone, with observed_at. A fact it cannot read is absent and named in unknown with why, never zeroed. Network filesystems are left out (statfs on a dead server blocks), and the statfs pass is bounded by the probe timeout.
  • The public uplink address is said as unknown with why: a machine behind a NAT cannot see it from itself, and how it learns it is not decided yet (see the report on issue 160).
  • cmd/mesh-host/main.go: the reconcile says its report unasked when the mesh last heard the profile 23 hours ago or more (machineEvery), so a machine nobody pushed to still says what it is once a day. Every heard report carrying a profile, a delivery's included, starts the day again.

Why

Issue 160: the mesh knew a machine's architecture and kernel family and nothing it is made of, and heard even that only after an apply. The decision on the issue (2026-10-11) names the facts, the daily refresh and that node shows them; the controller half is mesh-controller's PR for issue 160.

Tested

  • internal/profile/machine_test.go: every fact parsed from given files; a machine where nothing reads says every fact unknown with why and nothing as a value (the issue's acceptance); a hung statfs cannot hang the report; cores unsaid stay unknown; and against this machine (g14), the facts agree with what the process sees.
  • cmd/mesh-host/machine_refresh_test.go: an unchanged reconcile is quiet for 23 hours and speaks past it; a delivery heard restarts the day; a refusal is never due. The daily test fails with the check in news taken out.
  • sh merge-check.sh: pass (gofmt, vet, full suite with the race detector).

Fixes issue 160 (with the mesh-controller PR).

## What - `internal/profile/machine.go`: the profile now carries `machine`, what the machine is made of: memory, each local filesystem (mount, device, type, size, free), processor model, cores and threads, distribution and version, kernel release, virtual or physical (`systemd-detect-virt`), boot time (uptime is the time since) and timezone, with `observed_at`. A fact it cannot read is absent and named in `unknown` with why, never zeroed. Network filesystems are left out (statfs on a dead server blocks), and the statfs pass is bounded by the probe timeout. - The public uplink address is said as unknown with why: a machine behind a NAT cannot see it from itself, and how it learns it is not decided yet (see the report on issue 160). - `cmd/mesh-host/main.go`: the reconcile says its report unasked when the mesh last heard the profile 23 hours ago or more (`machineEvery`), so a machine nobody pushed to still says what it is once a day. Every heard report carrying a profile, a delivery's included, starts the day again. ## Why Issue 160: the mesh knew a machine's architecture and kernel family and nothing it is made of, and heard even that only after an apply. The decision on the issue (2026-10-11) names the facts, the daily refresh and that `node` shows them; the controller half is mesh-controller's PR for issue 160. ## Tested - `internal/profile/machine_test.go`: every fact parsed from given files; a machine where nothing reads says every fact unknown with why and nothing as a value (the issue's acceptance); a hung statfs cannot hang the report; cores unsaid stay unknown; and against this machine (g14), the facts agree with what the process sees. - `cmd/mesh-host/machine_refresh_test.go`: an unchanged reconcile is quiet for 23 hours and speaks past it; a delivery heard restarts the day; a refusal is never due. The daily test fails with the check in `news` taken out. - `sh merge-check.sh`: pass (gofmt, vet, full suite with the race detector). Fixes issue 160 (with the mesh-controller PR).
mesh-admin added 1 commit 2026-10-12 15:21:49 +00:00
Have a machine say what it is made of, and say it once a day unasked (issue 160)
mesh/merge-gate pass: builds mesh-host → ace, g14, novox, shanks; no bus step; every machine composes with the change as it did without (4 of 4 compose)
mesh/repo-check pass: its merge-check.sh passed
mesh/delivery superseded: a newer head of the same pull request
ec983bfc3d
The mesh knew a machine's architecture and kernel family and nothing it is made of, and heard even
that only when it sent the machine something. The profile now carries the machine's memory, each local
filesystem's size and free space, processor model, cores and threads, distribution and version, kernel
release, virtual or physical, boot time and timezone; a fact it cannot read is named unknown with why,
never zeroed. The reconcile says the profile unasked once the mesh last heard it 23 hours ago, so on a
machine nobody pushed to nothing the mesh holds of it is older than a day.
Author
Contributor

Delivery novox/mesh-host@c7a4bf5c11d6 — delivered since 2026-10-12T21:11:20Z

Delivery plan — builds mesh-host → ace, g14, novox, shanks; no bus step

  • tier 0: mesh-host
  • ace: receives mesh-host
  • g14: receives mesh-host
  • novox: receives mesh-host
  • shanks: receives mesh-host
  • grants step: if this changes what the bus's user list says (a new tool, subject or user), novox is sent that list alone, every build there kept, before mesh-host is judged on its first machine

Machines

  • mesh-host on ace: passed — healthy 3 times over 2m38s
  • mesh-host on the rest running it: sent

Transitions

  • 2026-10-12 15:42 (new) → proposed (announced): novox/mesh-host#92's head announced
  • 2026-10-12 15:48 proposed → checked (checked): the verdict names this commit
  • 2026-10-12 15:48 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
  • 2026-10-12 21:07 ready → published (merged): on the trunk its modules follow: merged there, and its walk opened — or nothing for a walk to move
  • 2026-10-12 21:07 published → delivering (go): its walk started: let go by this owner in its group's order, by a person, or on the controller's own path — not a failed walk that carried a later commit for it, after which it waits for a walk of its own commit
  • 2026-10-12 21:11 delivering → delivered (done): its walk is done, carrying its own commit: every machine of its deploy plan passed, or was left as its policy says

The commit's note under refs/notes/mesh-plan keeps every transition: git log --notes=mesh-plan.

<!-- mesh-delivery:view --> **Delivery** `novox/mesh-host@c7a4bf5c11d6` — **delivered** since 2026-10-12T21:11:20Z **Delivery plan** — builds mesh-host → ace, g14, novox, shanks; no bus step - tier 0: mesh-host - ace: receives mesh-host - g14: receives mesh-host - novox: receives mesh-host - shanks: receives mesh-host - grants step: if this changes what the bus's user list says (a new tool, subject or user), novox is sent that list alone, every build there kept, before mesh-host is judged on its first machine **Machines** - mesh-host on ace: passed — healthy 3 times over 2m38s - mesh-host on the rest running it: sent **Transitions** - 2026-10-12 15:42 (new) → proposed (announced): novox/mesh-host#92's head announced - 2026-10-12 15:48 proposed → checked (checked): the verdict names this commit - 2026-10-12 15:48 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail - 2026-10-12 21:07 ready → published (merged): on the trunk its modules follow: merged there, and its walk opened — or nothing for a walk to move - 2026-10-12 21:07 published → delivering (go): its walk started: let go by this owner in its group's order, by a person, or on the controller's own path — not a failed walk that carried a later commit for it, after which it waits for a walk of its own commit - 2026-10-12 21:11 delivering → delivered (done): its walk is done, carrying its own commit: every machine of its deploy plan passed, or was left as its policy says The commit's note under `refs/notes/mesh-plan` keeps every transition: `git log --notes=mesh-plan`.
Author
Contributor

Merge check at ec983bfc

  • mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: mesh-host)
  • mesh/repo-check: PASS — its merge-check.sh passed

Delivery plan — builds mesh-host → ace, g14, novox, shanks; no bus step

  • tier 0: mesh-host
  • ace: receives mesh-host
  • g14: receives mesh-host
  • novox: receives mesh-host
  • shanks: receives mesh-host
  • grants step: if this changes what the bus's user list says (a new tool, subject or user), novox is sent that list alone, every build there kept, before mesh-host is judged on its first machine
  • read by no module's build: cmd/mesh-host/machine_refresh_test.go, internal/profile/machine_test.go

Run by the build seat on novox as build-1791818523600674715 (builds --log build-1791818523600674715).

--- the gate: mesh-host
mesh-host: ok
long-running resources without health: 0
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-12T14:54:57Z says they run
composing every machine as the mesh is
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
composing every machine with the change
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
merge gate: PASS — every machine composes with the change as it did without (4 of 4 compose)
judged by controller development build against the facts of 2026-10-12T14:54:57Z

machines:
  - a machine (f35): composes
  - holds mesh-dns-resolver (idt): composes
  - holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
  - a machine (yvbpsm): composes

a merge would rebuild 1 module(s) in 1 tier(s): mesh-host

notes:
  - cmd/mesh-host/machine_refresh_test.go and internal/profile/machine_test.go read by no module's build, and rebuild nothing (issue 280)
--- its merge-check.sh (go toolchain)
ok  	github.com/novox/mesh-host/cmd/mesh-bootstrap	1.016s
ok  	github.com/novox/mesh-host/cmd/mesh-host	8.841s
ok  	github.com/novox/mesh-host/examples	1.020s
ok  	github.com/novox/mesh-host/internal/accounts	2.740s
ok  	github.com/novox/mesh-host/internal/apply	30.958s
ok  	github.com/novox/mesh-host/internal/bootstrap	3.507s
ok  	github.com/novox/mesh-host/internal/bundle	1.030s
ok  	github.com/novox/mesh-host/internal/declaration	1.099s
ok  	github.com/novox/mesh-host/internal/firewall	2.033s
ok  	github.com/novox/mesh-host/internal/identity	1.584s
ok  	github.com/novox/mesh-host/internal/image	1.022s
ok  	github.com/novox/mesh-host/internal/inventory	1.038s
ok  	github.com/novox/mesh-host/internal/link	10.551s
ok  	github.com/novox/mesh-host/internal/liveness	2.862s
ok  	github.com/novox/mesh-host/internal/meshcli	1.567s
ok  	github.com/novox/mesh-host/internal/network	1.044s
ok  	github.com/novox/mesh-host/internal/outward	1.021s
ok  	github.com/novox/mesh-host/internal/profile	1.562s
ok  	github.com/novox/mesh-host/internal/reachable	1.017s
ok  	github.com/novox/mesh-host/internal/store	2.225s
ok  	github.com/novox/mesh-host/internal/system	1.017s
ok  	github.com/novox/mesh-host/internal/tunnel	1.020s
ok  	github.com/novox/mesh-host/internal/units	1.011s
ok  	github.com/novox/mesh-host/internal/upgrade	2.112s
ok  	github.com/novox/mesh-host/internal/witness	36.116s
ok  	github.com/novox/mesh-host/packaging	1.014s
?   	github.com/novox/mesh-host/rootsearch	[no test files]
ok  	github.com/novox/mesh-host/validate	1.012s
**Merge check** at `ec983bfc` - `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: mesh-host) - `mesh/repo-check`: **PASS** — its merge-check.sh passed **Delivery plan** — builds mesh-host → ace, g14, novox, shanks; no bus step - tier 0: mesh-host - ace: receives mesh-host - g14: receives mesh-host - novox: receives mesh-host - shanks: receives mesh-host - grants step: if this changes what the bus's user list says (a new tool, subject or user), novox is sent that list alone, every build there kept, before mesh-host is judged on its first machine - read by no module's build: cmd/mesh-host/machine_refresh_test.go, internal/profile/machine_test.go Run by the build seat on novox as `build-1791818523600674715` (`builds --log build-1791818523600674715`). ``` --- the gate: mesh-host mesh-host: ok long-running resources without health: 0 files asking for a setting without saying whether it is trusted: 0 1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-12T14:54:57Z says they run composing every machine as the mesh is the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued composing every machine with the change the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued merge gate: PASS — every machine composes with the change as it did without (4 of 4 compose) judged by controller development build against the facts of 2026-10-12T14:54:57Z machines: - a machine (f35): composes - holds mesh-dns-resolver (idt): composes - holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes - a machine (yvbpsm): composes a merge would rebuild 1 module(s) in 1 tier(s): mesh-host notes: - cmd/mesh-host/machine_refresh_test.go and internal/profile/machine_test.go read by no module's build, and rebuild nothing (issue 280) --- its merge-check.sh (go toolchain) ok github.com/novox/mesh-host/cmd/mesh-bootstrap 1.016s ok github.com/novox/mesh-host/cmd/mesh-host 8.841s ok github.com/novox/mesh-host/examples 1.020s ok github.com/novox/mesh-host/internal/accounts 2.740s ok github.com/novox/mesh-host/internal/apply 30.958s ok github.com/novox/mesh-host/internal/bootstrap 3.507s ok github.com/novox/mesh-host/internal/bundle 1.030s ok github.com/novox/mesh-host/internal/declaration 1.099s ok github.com/novox/mesh-host/internal/firewall 2.033s ok github.com/novox/mesh-host/internal/identity 1.584s ok github.com/novox/mesh-host/internal/image 1.022s ok github.com/novox/mesh-host/internal/inventory 1.038s ok github.com/novox/mesh-host/internal/link 10.551s ok github.com/novox/mesh-host/internal/liveness 2.862s ok github.com/novox/mesh-host/internal/meshcli 1.567s ok github.com/novox/mesh-host/internal/network 1.044s ok github.com/novox/mesh-host/internal/outward 1.021s ok github.com/novox/mesh-host/internal/profile 1.562s ok github.com/novox/mesh-host/internal/reachable 1.017s ok github.com/novox/mesh-host/internal/store 2.225s ok github.com/novox/mesh-host/internal/system 1.017s ok github.com/novox/mesh-host/internal/tunnel 1.020s ok github.com/novox/mesh-host/internal/units 1.011s ok github.com/novox/mesh-host/internal/upgrade 2.112s ok github.com/novox/mesh-host/internal/witness 36.116s ok github.com/novox/mesh-host/packaging 1.014s ? github.com/novox/mesh-host/rootsearch [no test files] ok github.com/novox/mesh-host/validate 1.012s ```
mesh-admin left a comment
Author
Contributor

Verdict: request changes at ec983bfc3d

Independent review for issue 160 (a machine says little about itself).

Blocking

  1. The daily report does not reach the mesh within a day on a machine that sleeps. machineDue compares w.clock().Sub(w.machineSaid), both from time.Now(), so Go uses their monotonic readings. On Linux the runtime reads CLOCK_MONOTONIC (runtime/sys_linux_amd64.s, clock id 1), which does not advance while the machine is suspended. On the laptop (awake perhaps 8 hours a day), 23 monotonic hours can be about three calendar days, so the issue's acceptance ("the newest of it is no older than a day on a machine nobody has pushed to") fails on exactly the node that travels and sleeps. Fix: judge by wall time, for example store w.machineSaid = w.clock().Round(0) (which strips the monotonic reading) or compare .Round(0) values, and add a test where the said time carries no monotonic reading. A wall clock that jumps forward only causes one extra report, and a backward NTP step only delays one, so wall time is the safer judge here.

Checked and fine

  • Facts against the 2026-10-11 decision note: memory, each local filesystem with size and free, processor model, cores and threads, distribution and version, virtual or physical, uptime (as boot time), timezone and kernel release are all present. Each unreadable fact is omitted and named in unknown with why, never zeroed (TestMachine_aFactItCannotReadIsSaidUnknownNeverZero). Against g14 the output is correct: 8 cores, 16 threads, 30.6 GiB, physical, Europe/Brussels, /, /boot and the incus btrfs loop.
  • The public uplink address deviates from the note. The note says the node-engine adds it, and issue 238 depends on it. This PR always says it as unknown ("how it learns it is not decided yet"). That is honest under the unknown-with-why rule, so I do not block on it. Issue 160 must not be resolved on these two PRs, though: the public address, and the meshboard machine page the note also names, need their own issue (issue-tracker.open), per the finish-an-issue rule.
  • Bounded and not hanging: network filesystems are excluded by an allow-list of local types (nfs, cifs, sshfs and overlay are out), and the statfs pass is bounded by the probe timeout (10 s by default). Non-blocking: a statfs that truly hangs leaks one goroutine plus its OS thread per apply, and the whole filesystem list then reads unknown. Consider remembering a mount whose statfs is still outstanding and skipping it. fuseblk is in the list and is the only FUSE type that could hang that way.
  • Nothing sensitive leaks: only the cpuinfo model name is kept (no serials or flags), and no NFS server name is sent because network filesystems are excluded. Device paths and mount points do go on the bus. I accept that, since the bus is the mesh's own.
  • No report storm: the timer is local to the node-engine and resets on Heard, which fires when the broker takes the report, whatever the controller does with it. A controller restart or an older controller therefore cannot make it fire every 5 minutes. A simulation of 3 days of 5-minute reconciles, each heard, with observed_at changing every time, gave 4 reports (the start, then one every 23h05m). While unheard (link down or broker refusing), the report is due every reconcile, which is the existing retry semantics for unasked reports. A refusal is never due.
  • Backward compatible: machine is an added key with omitempty. The controller stores report.Profile whole as JSON, and its other profile readers (profileFrom for capabilities) ignore extra keys. Profile changes do not drive the controller's news (that comes from the doing record), so a changing observed_at cannot cause a push loop.
  • Tests fail without the fix: with adoptionWatch given only a now field and machineDue returning false on main, TestAMachineNobodyPushedToSaysWhatItIsOnceADay fails ("a day on, the reconcile did not say what the machine is").
  • sh merge-check.sh locally at this head: pass (gofmt, vet, full suite under -race). There is no nats-server here, so the bus tests were skipped locally.
  • Statuses at this head: mesh/merge-gate success (4 of 4 compose), mesh/repo-check success, mesh/delivery success (ready).
Verdict: request changes at ec983bfc3d1b52443f08095f611c19e04c8d201c Independent review for issue 160 (a machine says little about itself). ## Blocking 1. **The daily report does not reach the mesh within a day on a machine that sleeps.** `machineDue` compares `w.clock().Sub(w.machineSaid)`, both from `time.Now()`, so Go uses their monotonic readings. On Linux the runtime reads CLOCK_MONOTONIC (`runtime/sys_linux_amd64.s`, clock id 1), which does not advance while the machine is suspended. On the laptop (awake perhaps 8 hours a day), 23 monotonic hours can be about three calendar days, so the issue's acceptance ("the newest of it is no older than a day on a machine nobody has pushed to") fails on exactly the node that travels and sleeps. Fix: judge by wall time, for example store `w.machineSaid = w.clock().Round(0)` (which strips the monotonic reading) or compare `.Round(0)` values, and add a test where the said time carries no monotonic reading. A wall clock that jumps forward only causes one extra report, and a backward NTP step only delays one, so wall time is the safer judge here. ## Checked and fine - **Facts against the 2026-10-11 decision note:** memory, each local filesystem with size and free, processor model, cores and threads, distribution and version, virtual or physical, uptime (as boot time), timezone and kernel release are all present. Each unreadable fact is omitted and named in `unknown` with why, never zeroed (`TestMachine_aFactItCannotReadIsSaidUnknownNeverZero`). Against g14 the output is correct: 8 cores, 16 threads, 30.6 GiB, physical, Europe/Brussels, `/`, `/boot` and the incus btrfs loop. - **The public uplink address deviates from the note.** The note says the node-engine adds it, and issue 238 depends on it. This PR always says it as unknown ("how it learns it is not decided yet"). That is honest under the unknown-with-why rule, so I do not block on it. Issue 160 must not be resolved on these two PRs, though: the public address, and the meshboard machine page the note also names, need their own issue (`issue-tracker.open`), per the finish-an-issue rule. - **Bounded and not hanging:** network filesystems are excluded by an allow-list of local types (nfs, cifs, sshfs and overlay are out), and the statfs pass is bounded by the probe timeout (10 s by default). Non-blocking: a statfs that truly hangs leaks one goroutine plus its OS thread per apply, and the whole filesystem list then reads unknown. Consider remembering a mount whose statfs is still outstanding and skipping it. `fuseblk` is in the list and is the only FUSE type that could hang that way. - **Nothing sensitive leaks:** only the cpuinfo model name is kept (no serials or flags), and no NFS server name is sent because network filesystems are excluded. Device paths and mount points do go on the bus. I accept that, since the bus is the mesh's own. - **No report storm:** the timer is local to the node-engine and resets on `Heard`, which fires when the broker takes the report, whatever the controller does with it. A controller restart or an older controller therefore cannot make it fire every 5 minutes. A simulation of 3 days of 5-minute reconciles, each heard, with `observed_at` changing every time, gave 4 reports (the start, then one every 23h05m). While unheard (link down or broker refusing), the report is due every reconcile, which is the existing retry semantics for unasked reports. A refusal is never due. - **Backward compatible:** `machine` is an added key with `omitempty`. The controller stores `report.Profile` whole as JSON, and its other profile readers (`profileFrom` for capabilities) ignore extra keys. Profile changes do not drive the controller's `news` (that comes from the doing record), so a changing `observed_at` cannot cause a push loop. - **Tests fail without the fix:** with `adoptionWatch` given only a `now` field and `machineDue` returning false on main, `TestAMachineNobodyPushedToSaysWhatItIsOnceADay` fails ("a day on, the reconcile did not say what the machine is"). - `sh merge-check.sh` locally at this head: pass (gofmt, vet, full suite under -race). There is no nats-server here, so the bus tests were skipped locally. - Statuses at this head: mesh/merge-gate success (4 of 4 compose), mesh/repo-check success, mesh/delivery success (ready).
jschoubben added 1 commit 2026-10-12 15:42:21 +00:00
Judge the daily report by wall time, and keep one filesystem check at a time (issue 160)
mesh/merge-gate head c7a4bf5c11d6 merged into trunk d8e785bb4b4b — pass: builds mesh-host → ace, g14, novox, shanks; no bus step; every machine composes wi…
mesh/repo-check head c7a4bf5c11d6 merged into trunk d8e785bb4b4b — pass: its merge-check.sh passed
mesh/delivery delivered
c7a4bf5c11
Go's monotonic clock stops while a laptop is suspended, so a day judged by it could pass a day and a half
of wall time before the machine said what it is. And a hung statfs left a goroutine and thread blocked per
apply; now one check runs at a time and reports after it say the sizes unknown until it ends.
Author
Contributor

Merge check at c7a4bf5c

  • mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: mesh-host)
  • mesh/repo-check: PASS — its merge-check.sh passed

Delivery plan — builds mesh-host → ace, g14, novox, shanks; no bus step

  • tier 0: mesh-host
  • ace: receives mesh-host
  • g14: receives mesh-host
  • novox: receives mesh-host
  • shanks: receives mesh-host
  • grants step: if this changes what the bus's user list says (a new tool, subject or user), novox is sent that list alone, every build there kept, before mesh-host is judged on its first machine
  • read by no module's build: cmd/mesh-host/machine_refresh_test.go, internal/profile/machine_test.go

Run by the build seat on g14 as build-1791819771323743381 (builds --log build-1791819771323743381).

--- the gate: mesh-host
mesh-host: ok
long-running resources without health: 0
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-12T15:38:39Z says they run
composing every machine as the mesh is
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
composing every machine with the change
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
merge gate: PASS — every machine composes with the change as it did without (4 of 4 compose)
judged by controller development build against the facts of 2026-10-12T15:38:39Z

machines:
  - a machine (f35): composes
  - holds mesh-dns-resolver (idt): composes
  - holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
  - a machine (yvbpsm): composes

a merge would rebuild 1 module(s) in 1 tier(s): mesh-host

notes:
  - cmd/mesh-host/machine_refresh_test.go and internal/profile/machine_test.go read by no module's build, and rebuild nothing (issue 280)
--- its merge-check.sh (go toolchain)
ok  	github.com/novox/mesh-host/cmd/mesh-bootstrap	1.009s
ok  	github.com/novox/mesh-host/cmd/mesh-host	5.632s
ok  	github.com/novox/mesh-host/examples	1.014s
ok  	github.com/novox/mesh-host/internal/accounts	1.507s
ok  	github.com/novox/mesh-host/internal/apply	2.909s
ok  	github.com/novox/mesh-host/internal/bootstrap	1.521s
ok  	github.com/novox/mesh-host/internal/bundle	1.019s
ok  	github.com/novox/mesh-host/internal/declaration	1.054s
ok  	github.com/novox/mesh-host/internal/firewall	1.680s
ok  	github.com/novox/mesh-host/internal/identity	1.092s
ok  	github.com/novox/mesh-host/internal/image	1.019s
ok  	github.com/novox/mesh-host/internal/inventory	1.029s
ok  	github.com/novox/mesh-host/internal/link	10.496s
ok  	github.com/novox/mesh-host/internal/liveness	2.061s
ok  	github.com/novox/mesh-host/internal/meshcli	1.570s
ok  	github.com/novox/mesh-host/internal/network	1.033s
ok  	github.com/novox/mesh-host/internal/outward	1.024s
ok  	github.com/novox/mesh-host/internal/profile	1.596s
ok  	github.com/novox/mesh-host/internal/reachable	1.023s
ok  	github.com/novox/mesh-host/internal/store	1.081s
ok  	github.com/novox/mesh-host/internal/system	1.011s
ok  	github.com/novox/mesh-host/internal/tunnel	1.014s
ok  	github.com/novox/mesh-host/internal/units	1.008s
ok  	github.com/novox/mesh-host/internal/upgrade	1.043s
ok  	github.com/novox/mesh-host/internal/witness	1.539s
ok  	github.com/novox/mesh-host/packaging	1.007s
?   	github.com/novox/mesh-host/rootsearch	[no test files]
ok  	github.com/novox/mesh-host/validate	1.007s
**Merge check** at `c7a4bf5c` - `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: mesh-host) - `mesh/repo-check`: **PASS** — its merge-check.sh passed **Delivery plan** — builds mesh-host → ace, g14, novox, shanks; no bus step - tier 0: mesh-host - ace: receives mesh-host - g14: receives mesh-host - novox: receives mesh-host - shanks: receives mesh-host - grants step: if this changes what the bus's user list says (a new tool, subject or user), novox is sent that list alone, every build there kept, before mesh-host is judged on its first machine - read by no module's build: cmd/mesh-host/machine_refresh_test.go, internal/profile/machine_test.go Run by the build seat on g14 as `build-1791819771323743381` (`builds --log build-1791819771323743381`). ``` --- the gate: mesh-host mesh-host: ok long-running resources without health: 0 files asking for a setting without saying whether it is trusted: 0 1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-12T15:38:39Z says they run composing every machine as the mesh is the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued composing every machine with the change the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued merge gate: PASS — every machine composes with the change as it did without (4 of 4 compose) judged by controller development build against the facts of 2026-10-12T15:38:39Z machines: - a machine (f35): composes - holds mesh-dns-resolver (idt): composes - holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes - a machine (yvbpsm): composes a merge would rebuild 1 module(s) in 1 tier(s): mesh-host notes: - cmd/mesh-host/machine_refresh_test.go and internal/profile/machine_test.go read by no module's build, and rebuild nothing (issue 280) --- its merge-check.sh (go toolchain) ok github.com/novox/mesh-host/cmd/mesh-bootstrap 1.009s ok github.com/novox/mesh-host/cmd/mesh-host 5.632s ok github.com/novox/mesh-host/examples 1.014s ok github.com/novox/mesh-host/internal/accounts 1.507s ok github.com/novox/mesh-host/internal/apply 2.909s ok github.com/novox/mesh-host/internal/bootstrap 1.521s ok github.com/novox/mesh-host/internal/bundle 1.019s ok github.com/novox/mesh-host/internal/declaration 1.054s ok github.com/novox/mesh-host/internal/firewall 1.680s ok github.com/novox/mesh-host/internal/identity 1.092s ok github.com/novox/mesh-host/internal/image 1.019s ok github.com/novox/mesh-host/internal/inventory 1.029s ok github.com/novox/mesh-host/internal/link 10.496s ok github.com/novox/mesh-host/internal/liveness 2.061s ok github.com/novox/mesh-host/internal/meshcli 1.570s ok github.com/novox/mesh-host/internal/network 1.033s ok github.com/novox/mesh-host/internal/outward 1.024s ok github.com/novox/mesh-host/internal/profile 1.596s ok github.com/novox/mesh-host/internal/reachable 1.023s ok github.com/novox/mesh-host/internal/store 1.081s ok github.com/novox/mesh-host/internal/system 1.011s ok github.com/novox/mesh-host/internal/tunnel 1.014s ok github.com/novox/mesh-host/internal/units 1.008s ok github.com/novox/mesh-host/internal/upgrade 1.043s ok github.com/novox/mesh-host/internal/witness 1.539s ok github.com/novox/mesh-host/packaging 1.007s ? github.com/novox/mesh-host/rootsearch [no test files] ok github.com/novox/mesh-host/validate 1.007s ```
mesh-admin left a comment
Author
Contributor

Verdict: approve at c7a4bf5c11

This re-review checks the head against my review 84 (at ec983bf). I read only the change since then: c7a4bf5.

  • The blocking point is fixed. clock() now returns .Round(0) for both the given clock and time.Now(), so the stored machineSaid and the compared time are both wall time. A suspended laptop's sleep now counts toward the day.
  • The new clock test fails without the fix. I put the new machine_refresh_test.go on the old code (ec983bf). TestTheWatchClockIsWallTime fails there, because the clock carries a monotonic reading (m=+0.0008).
  • The goroutine leak is fixed. The filesystem check takes a package-wide sizing.TryLock(), and the goroutine releases it when its statfs finally returns. While a check hangs, every later report says right away "has not finished yet" and asks no filesystem again. The lock cannot be held forever: if statfs never returns, sizes simply stay unknown with that why, which is the honest answer.
  • The new filesystem test fails without the fix. With the new hung-filesystem test on the old code (its cleanup loop adapted), it fails: "filesystems asked 4 times while one check hung, was 1".
  • sh merge-check.sh at this head in my clone: exit 0 (gofmt, vet, full suite under -race). There is no nats-server here, so the bus tests were skipped locally.
  • Statuses at this head: mesh/merge-gate success (4 of 4 compose), mesh/repo-check success, mesh/delivery success (ready).

Still standing from review 84, and not blocking this PR: the public uplink address is always reported as unknown, while the 2026-10-11 decision says the node-engine reports it. The meshboard page is not built either. Both need their own issue before issue 160 is resolved.

Verdict: approve at c7a4bf5c11d6201af3797aa80fe38794f1624f83 This re-review checks the head against my review 84 (at ec983bf). I read only the change since then: c7a4bf5. - **The blocking point is fixed.** `clock()` now returns `.Round(0)` for both the given clock and `time.Now()`, so the stored `machineSaid` and the compared time are both wall time. A suspended laptop's sleep now counts toward the day. - **The new clock test fails without the fix.** I put the new `machine_refresh_test.go` on the old code (ec983bf). `TestTheWatchClockIsWallTime` fails there, because the clock carries a monotonic reading (`m=+0.0008`). - **The goroutine leak is fixed.** The filesystem check takes a package-wide `sizing.TryLock()`, and the goroutine releases it when its statfs finally returns. While a check hangs, every later report says right away "has not finished yet" and asks no filesystem again. The lock cannot be held forever: if statfs never returns, sizes simply stay unknown with that why, which is the honest answer. - **The new filesystem test fails without the fix.** With the new hung-filesystem test on the old code (its cleanup loop adapted), it fails: "filesystems asked 4 times while one check hung, was 1". - `sh merge-check.sh` at this head in my clone: exit 0 (gofmt, vet, full suite under -race). There is no nats-server here, so the bus tests were skipped locally. - Statuses at this head: mesh/merge-gate success (4 of 4 compose), mesh/repo-check success, mesh/delivery success (ready). Still standing from review 84, and not blocking this PR: the public uplink address is always reported as unknown, while the 2026-10-11 decision says the node-engine reports it. The meshboard page is not built either. Both need their own issue before issue 160 is resolved.
Author
Contributor

Merge check at c7a4bf5c
Judged: head c7a4bf5c11d6 merged into trunk d8e785bb4b4b; read beside it: mesh-catalog 9c9a76d220f8, mesh-controller 2f157148c6c8, mesh-controller-main 2f157148c6c8, mesh-tools 954e1a53eef9

  • mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: mesh-host)
  • mesh/repo-check: PASS — its merge-check.sh passed

Delivery plan — builds mesh-host → ace, g14, novox, shanks; no bus step

  • tier 0: mesh-host
  • ace: receives mesh-host
  • g14: receives mesh-host
  • novox: receives mesh-host
  • shanks: receives mesh-host
  • grants step: if this changes what the bus's user list says (a new tool, subject or user), novox is sent that list alone, every build there kept, before mesh-host is judged on its first machine
  • read by no module's build: cmd/mesh-host/machine_refresh_test.go, internal/profile/machine_test.go

Run by the build seat on ace as build-1791838287474578783 (builds --log build-1791838287474578783).

--- the gate: mesh-host
mesh-host: ok
long-running resources without health: 0
files asking for a setting without saying whether it is trusted: 0
1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here
judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-12T20:48:25Z says they run
composing every machine as the mesh is
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
composing every machine with the change
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued
merge gate: PASS — every machine composes with the change as it did without (4 of 4 compose)
judged by controller development build against the facts of 2026-10-12T20:48:25Z

machines:
  - a machine (f35): composes
  - holds mesh-dns-resolver (idt): composes
  - holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes
  - a machine (yvbpsm): composes

a merge would rebuild 1 module(s) in 1 tier(s): mesh-host

notes:
  - cmd/mesh-host/machine_refresh_test.go and internal/profile/machine_test.go read by no module's build, and rebuild nothing (issue 280)
--- its merge-check.sh (go toolchain)
ok  	github.com/novox/mesh-host/cmd/mesh-bootstrap	1.014s
ok  	github.com/novox/mesh-host/cmd/mesh-host	5.527s
ok  	github.com/novox/mesh-host/examples	1.018s
ok  	github.com/novox/mesh-host/internal/accounts	1.407s
ok  	github.com/novox/mesh-host/internal/apply	3.225s
ok  	github.com/novox/mesh-host/internal/bootstrap	1.476s
ok  	github.com/novox/mesh-host/internal/bundle	1.034s
ok  	github.com/novox/mesh-host/internal/declaration	1.070s
ok  	github.com/novox/mesh-host/internal/firewall	1.657s
ok  	github.com/novox/mesh-host/internal/identity	1.073s
ok  	github.com/novox/mesh-host/internal/image	1.036s
ok  	github.com/novox/mesh-host/internal/inventory	1.045s
ok  	github.com/novox/mesh-host/internal/link	10.495s
ok  	github.com/novox/mesh-host/internal/liveness	2.099s
ok  	github.com/novox/mesh-host/internal/meshcli	1.570s
ok  	github.com/novox/mesh-host/internal/network	1.051s
ok  	github.com/novox/mesh-host/internal/outward	1.044s
ok  	github.com/novox/mesh-host/internal/profile	1.585s
ok  	github.com/novox/mesh-host/internal/reachable	1.040s
ok  	github.com/novox/mesh-host/internal/store	1.156s
ok  	github.com/novox/mesh-host/internal/system	1.042s
ok  	github.com/novox/mesh-host/internal/tunnel	1.060s
ok  	github.com/novox/mesh-host/internal/units	1.053s
ok  	github.com/novox/mesh-host/internal/upgrade	1.093s
ok  	github.com/novox/mesh-host/internal/witness	2.552s
ok  	github.com/novox/mesh-host/packaging	1.033s
?   	github.com/novox/mesh-host/rootsearch	[no test files]
ok  	github.com/novox/mesh-host/validate	1.036s
**Merge check** at `c7a4bf5c` Judged: head `c7a4bf5c11d6` merged into trunk `d8e785bb4b4b`; read beside it: mesh-catalog `9c9a76d220f8`, mesh-controller `2f157148c6c8`, mesh-controller-main `2f157148c6c8`, mesh-tools `954e1a53eef9` - `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: mesh-host) - `mesh/repo-check`: **PASS** — its merge-check.sh passed **Delivery plan** — builds mesh-host → ace, g14, novox, shanks; no bus step - tier 0: mesh-host - ace: receives mesh-host - g14: receives mesh-host - novox: receives mesh-host - shanks: receives mesh-host - grants step: if this changes what the bus's user list says (a new tool, subject or user), novox is sent that list alone, every build there kept, before mesh-host is judged on its first machine - read by no module's build: cmd/mesh-host/machine_refresh_test.go, internal/profile/machine_test.go Run by the build seat on ace as `build-1791838287474578783` (`builds --log build-1791838287474578783`). ``` --- the gate: mesh-host mesh-host: ok long-running resources without health: 0 files asking for a setting without saying whether it is trusted: 0 1 manifest(s) checked. Judged against the seats this binary carries; a claim on one of the mesh's own seats is judged fully at registration, and a seat declared by a module not given here reads as unknown. Identities judged on a 6-character machine name, against the bounds of the providers given here judging 0 module(s) changed against 4 machine(s), as the snapshot of 2026-10-12T20:48:25Z says they run composing every machine as the mesh is the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued composing every machine with the change the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued the bus's user list leaves out 5 user(s) the mesh has minted no credential for: controller, node.f35, node.idt, node.ygcyr, node.yvbpsm. Each is a user that cannot connect until one is issued merge gate: PASS — every machine composes with the change as it did without (4 of 4 compose) judged by controller development build against the facts of 2026-10-12T20:48:25Z machines: - a machine (f35): composes - holds mesh-dns-resolver (idt): composes - holds git, holds mesh-artifact-store, holds mesh-broker, holds mesh-catalog, holds mesh-delivery, holds mesh-dns-resolver, holds mesh-store, holds mesh-vault, holds npm-package-registry, the control node, the hub (ygcyr): composes - a machine (yvbpsm): composes a merge would rebuild 1 module(s) in 1 tier(s): mesh-host notes: - cmd/mesh-host/machine_refresh_test.go and internal/profile/machine_test.go read by no module's build, and rebuild nothing (issue 280) --- its merge-check.sh (go toolchain) ok github.com/novox/mesh-host/cmd/mesh-bootstrap 1.014s ok github.com/novox/mesh-host/cmd/mesh-host 5.527s ok github.com/novox/mesh-host/examples 1.018s ok github.com/novox/mesh-host/internal/accounts 1.407s ok github.com/novox/mesh-host/internal/apply 3.225s ok github.com/novox/mesh-host/internal/bootstrap 1.476s ok github.com/novox/mesh-host/internal/bundle 1.034s ok github.com/novox/mesh-host/internal/declaration 1.070s ok github.com/novox/mesh-host/internal/firewall 1.657s ok github.com/novox/mesh-host/internal/identity 1.073s ok github.com/novox/mesh-host/internal/image 1.036s ok github.com/novox/mesh-host/internal/inventory 1.045s ok github.com/novox/mesh-host/internal/link 10.495s ok github.com/novox/mesh-host/internal/liveness 2.099s ok github.com/novox/mesh-host/internal/meshcli 1.570s ok github.com/novox/mesh-host/internal/network 1.051s ok github.com/novox/mesh-host/internal/outward 1.044s ok github.com/novox/mesh-host/internal/profile 1.585s ok github.com/novox/mesh-host/internal/reachable 1.040s ok github.com/novox/mesh-host/internal/store 1.156s ok github.com/novox/mesh-host/internal/system 1.042s ok github.com/novox/mesh-host/internal/tunnel 1.060s ok github.com/novox/mesh-host/internal/units 1.053s ok github.com/novox/mesh-host/internal/upgrade 1.093s ok github.com/novox/mesh-host/internal/witness 2.552s ok github.com/novox/mesh-host/packaging 1.033s ? github.com/novox/mesh-host/rootsearch [no test files] ok github.com/novox/mesh-host/validate 1.036s ```
mesh-admin merged commit 8e95c1d9ed into main 2026-10-12 21:07:05 +00:00
mesh-admin deleted branch fix/160-a-machine-says-what-it-is 2026-10-12 21:07:06 +00:00
Sign in to join this conversation.
No Reviewers
2 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: novox/mesh-host#92