This pull request makes the fixes from both safety reviews of the zfs module (merged as 90516fc), and adds the operator's requirement that the module owns the pool's scrub. Do not merge before mesh-controller #171. Do not assign before #171 is live. Do not unassign before mesh-host #66 (hq ADR 0269, proposed: hq #238) is merged.
The scrub schedule
Settings:
scrub-pool: the operator's value, with no default.
scrub-cadence: weekly or monthly, default weekly.
scrub-day: Mon to Sun, default Mon.
scrub-time: HH:MM, default 00:00.
The defaults match the distribution's weekly timer on the home server today. The operator chose monthly for ace.
The scheduled run. The node-engine runs the scheduled process zfs-mesh-scrub (zfs-tools scrub-run, as root) from its own zfs-mesh-scrub.timer every 15 minutes. The run finds the slot (the last scheduled time not after now). If that slot is not served yet:
if a scrub is running, it waits with zpool wait -t scrub -- <pool>;
otherwise it starts one with zpool scrub -- <pool>, records the slot as served, and waits;
if the start is refused with "currently scrubbing", it waits for that scrub;
a running resilver defers the scrub.
It never pauses or stops a scrub, and the unit has no ExecStop. The served slot is kept in /var/lib/zfs-tools/scrub-served.json.
No setting value reaches a unit file. The day, time and cadence go only into /etc/zfs-tools/scrub.conf. The run checks them against fixed sets and refuses anything else (*:* and Mon..Fri among them) without running anything. The pool name is the only setting in a unit name: the controller fills it into the distribution timers' names and refuses anything but [A-Za-z0-9:_.-] (mesh-controller #171).
Exactly one schedule. The distribution's zfs-scrub-weekly@<pool>.timer and zfs-scrub-monthly@<pool>.timer are held stopped and disabled. They come after the scheduled run in the declaration, so a run that cannot be installed stops the apply before today's timer is touched. Unassigning gives them back running and enabled, once mesh-host #66 is merged.
Tools.
zfs_scrub shows the schedule, the next scheduled scrub, the timer's state, any resilver's progress and time to go, and the last scrub and resilver.
zfs_scrub_start, zfs_scrub_pause and zfs_scrub_stop take {pool, why} and run sudo -n zpool scrub [-p|-s] -- <pool>. They are recorded as scrub.requested, and are candidates for the approval mark (research 039).
Health warnings:
a scrub later than the cadence plus a margin;
a scheduled scrub that has not started an hour after its time;
a failed run;
a timer that does not run;
invalid settings;
a running resilver.
A resilver that finished with errors is unhealthy.
Composed for ace by the controller from #171, with the real manifest, scrub-pool=storage and scrub-cadence=monthly:
zfs.scrub-run: process zfs-mesh-scrub, schedule */15 * * * * (units zfs-mesh-scrub.timer and .service)
zfs.distribution-weekly: zfs-scrub-weekly@storage.timer, stopped and disabled
zfs.distribution-monthly: zfs-scrub-monthly@storage.timer, stopped and disabled
zfs_snapshot: no recursive (passing it is refused). Every answer gives the dataset's free space and the pool's capacity, and says the snapshot holds every block deleted or overwritten until it is destroyed by hand. A timed-out command is answered from zfs list, or as "outcome unknown; check zfs_snapshots".
Timeouts: a command that runs out of time is ended with its whole process group (SIGTERM, which sudo passes on, then SIGKILL after 5 s).
The watch runs as the operator account.
The event follower:
it reads ZFS's queue from its start;
a per-boot cursor in /run/zfs-events keeps it from writing an event twice;
it writes "dropped N events";
it kills zpool before waiting for it;
it exits non-zero after 5 failed follows in a row.
The home server's journal is on disk, so the journaled events survive a reboot.
Health: error counts from before are a warning; counts that rise are unhealthy for 15 minutes. The counts are kept in /var/lib/zfs-tools/errors.json across restarts.
The guard test catches every way to run a command (ask, AsRoot, Escalated, .Run(, .Start(, .Output(, exec.Command(Context)(, os.StartProcess, syscall.Exec). It holds each argument list to its program and verb. The scheduled run may only use scrub (start) and wait.
Fixtures and parser follow OpenZFS 2.4.4's source: human scan sizes under -p, no iostat section lines, error scrubs, per-vdev rebuilds, paused scrubs.
Checked
Tests were written first. go test -race passes in UTC, UTC+14, UTC−12 and Europe/Brussels.
merge-check.sh with the controller from #171: zfs: ok, 15 tools, health checks declared, long-running resources without health 70 as kept.
At bd01827, mesh/merge-gate and mesh/repo-check pass.
This pull request makes the fixes from both safety reviews of the `zfs` module (merged as 90516fc), and adds the operator's requirement that the module owns the pool's scrub. **Do not merge before mesh-controller #171. Do not assign before #171 is live. Do not unassign before mesh-host #66 (hq ADR 0269, proposed: hq #238) is merged.**
## The scrub schedule
- **Settings:**
- `scrub-pool`: the operator's value, with no default.
- `scrub-cadence`: `weekly` or `monthly`, default `weekly`.
- `scrub-day`: `Mon` to `Sun`, default `Mon`.
- `scrub-time`: `HH:MM`, default `00:00`.
The defaults match the distribution's weekly timer on the home server today. The operator chose monthly for ace.
- **The scheduled run.** The node-engine runs the scheduled process `zfs-mesh-scrub` (`zfs-tools scrub-run`, as root) from its own `zfs-mesh-scrub.timer` every 15 minutes. The run finds the slot (the last scheduled time not after now). If that slot is not served yet:
- if a scrub is running, it waits with `zpool wait -t scrub -- <pool>`;
- otherwise it starts one with `zpool scrub -- <pool>`, records the slot as served, and waits;
- if the start is refused with "currently scrubbing", it waits for that scrub;
- a running resilver defers the scrub.
It never pauses or stops a scrub, and the unit has no ExecStop. The served slot is kept in `/var/lib/zfs-tools/scrub-served.json`.
- **No setting value reaches a unit file.** The day, time and cadence go only into `/etc/zfs-tools/scrub.conf`. The run checks them against fixed sets and refuses anything else (`*:*` and `Mon..Fri` among them) without running anything. The pool name is the only setting in a unit name: the controller fills it into the distribution timers' names and refuses anything but `[A-Za-z0-9:_.-]` (mesh-controller #171).
- **Exactly one schedule.** The distribution's `zfs-scrub-weekly@<pool>.timer` and `zfs-scrub-monthly@<pool>.timer` are held stopped and disabled. They come after the scheduled run in the declaration, so a run that cannot be installed stops the apply before today's timer is touched. Unassigning gives them back running and enabled, once mesh-host #66 is merged.
- **Tools.**
- `zfs_scrub` shows the schedule, the next scheduled scrub, the timer's state, any resilver's progress and time to go, and the last scrub and resilver.
- `zfs_scrub_start`, `zfs_scrub_pause` and `zfs_scrub_stop` take `{pool, why}` and run `sudo -n zpool scrub [-p|-s] -- <pool>`. They are recorded as `scrub.requested`, and are candidates for the approval mark (research 039).
- **Health warnings:**
- a scrub later than the cadence plus a margin;
- a scheduled scrub that has not started an hour after its time;
- a failed run;
- a timer that does not run;
- invalid settings;
- a running resilver.
A resilver that finished with errors is unhealthy.
**Composed for ace** by the controller from #171, with the real manifest, `scrub-pool=storage` and `scrub-cadence=monthly`:
- `zfs.scrub-run`: process `zfs-mesh-scrub`, schedule `*/15 * * * *` (units `zfs-mesh-scrub.timer` and `.service`)
- `zfs.distribution-weekly`: `zfs-scrub-weekly@storage.timer`, stopped and disabled
- `zfs.distribution-monthly`: `zfs-scrub-monthly@storage.timer`, stopped and disabled
- `zfs.scrub-config`: `pool=storage`, `cadence=monthly`, `day=Mon`, `time=00:00`
The next scrub would be 2026-11-02 00:00.
## The first review's fixes
- **`zfs_snapshot`:** no `recursive` (passing it is refused). Every answer gives the dataset's free space and the pool's capacity, and says the snapshot holds every block deleted or overwritten until it is destroyed by hand. A timed-out command is answered from `zfs list`, or as "outcome unknown; check zfs_snapshots".
- **Timeouts:** a command that runs out of time is ended with its whole process group (SIGTERM, which sudo passes on, then SIGKILL after 5 s).
- **The watch** runs as the operator account.
- **The event follower:**
- it reads ZFS's queue from its start;
- a per-boot cursor in `/run/zfs-events` keeps it from writing an event twice;
- it writes "dropped N events";
- it kills zpool before waiting for it;
- it exits non-zero after 5 failed follows in a row.
The home server's journal is on disk, so the journaled events survive a reboot.
- **Health:** error counts from before are a warning; counts that rise are unhealthy for 15 minutes. The counts are kept in `/var/lib/zfs-tools/errors.json` across restarts.
- **The guard test** catches every way to run a command (`ask`, `AsRoot`, `Escalated`, `.Run(`, `.Start(`, `.Output(`, `exec.Command(Context)(`, `os.StartProcess`, `syscall.Exec`). It holds each argument list to its program and verb. The scheduled run may only use `scrub` (start) and `wait`.
- **Fixtures and parser** follow OpenZFS 2.4.4's source: human scan sizes under `-p`, no iostat section lines, error scrubs, per-vdev rebuilds, paused scrubs.
## Checked
- Tests were written first. `go test -race` passes in UTC, UTC+14, UTC−12 and Europe/Brussels.
- `merge-check.sh` with the controller from #171: zfs: ok, 15 tools, health checks declared, long-running resources without health 70 as kept.
- At bd01827, `mesh/merge-gate` and `mesh/repo-check` pass.
The safety review of the first version found it safe for the pool's data and asked for these fixes before
it goes on the home server:
- zfs_snapshot: no recursive. Every answer gives the dataset's free space and the pool's capacity, and
says that the snapshot holds every block deleted or overwritten until it is destroyed by hand.
- A command that runs out of time is ended with its whole process group. An act that timed out is
answered from the list, or as "outcome unknown", never as "NOT DONE".
- The watch runs as the operator account.
- The event follower reads ZFS's queue from its start, with a cursor per boot, so the pool's import and
any downtime reach the journal once. It writes "dropped N events", kills zpool before waiting for it,
and exits non-zero after five failed follows in a row.
- Past error counts are a warning, and counts that rise are unhealthy.
- The guard test catches .Run(, exec.Command(, exec.CommandContext( and process starts.
- The fixtures and the parser follow OpenZFS 2.4.4's own printers: human scan sizes under -p, no iostat
section lines, error scrubs, per-vdev rebuild resilvers, paused scrubs.
The operator also asked the module to own the scrub. A timer and a service of the module's own run
zpool scrub -w -- <pool> from the scrub-pool, scrub-cadence, scrub-day and scrub-time settings. In the
same declaration, the distribution's weekly and monthly timers for that pool are held stopped and
disabled, so exactly one schedule scrubs it. zfs_scrub shows the schedule and the next run, with any
resilver's progress. zfs_scrub_start, zfs_scrub_pause and zfs_scrub_stop act with why. The health check
warns of a late or failed scrub and of a running resilver, and is unhealthy for a resilver that finished
with errors.
Deliverynovox/mesh-catalog@a9acd3dd289b — delivered since 2026-10-09T01:03:49Z
Delivery plan — builds zfs, sent nowhere; no bus step
tier 0: zfs
Machines
zfs on every machine running it: sent
Transitions
2026-10-08 23:55 (new) → proposed (announced): novox/mesh-catalog#147's head announced
2026-10-08 23:58 proposed → checked (checked): the verdict names this commit
2026-10-08 23:58 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
2026-10-09 00:54 ready → proposed (recheck): checked at 23:58Z before controller #170/#171 were live; the gate must judge its unit-name settings with the live controller before merging
2026-10-09 00:56 proposed → checked (checked): the verdict names this commit
2026-10-09 00:56 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
2026-10-09 01:02 ready → published (merged): on the trunk its modules follow: merged there, and its walk opened — or nothing for a walk to move
2026-10-09 01:03 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
2026-10-09 01:03 delivering → delivered (done): its walk is done: 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-catalog@a9acd3dd289b` — **delivered** since 2026-10-09T01:03:49Z
**Delivery plan** — builds zfs, sent nowhere; no bus step
- tier 0: zfs
**Machines**
- zfs on every machine running it: sent
**Transitions**
- 2026-10-08 23:55 (new) → proposed (announced): novox/mesh-catalog#147's head announced
- 2026-10-08 23:58 proposed → checked (checked): the verdict names this commit
- 2026-10-08 23:58 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
- 2026-10-09 00:54 ready → proposed (recheck): checked at 23:58Z before controller #170/#171 were live; the gate must judge its unit-name settings with the live controller before merging
- 2026-10-09 00:56 proposed → checked (checked): the verdict names this commit
- 2026-10-09 00:56 checked → ready (accepted): the gate passed or warned, and the repository's own check did not fail
- 2026-10-09 01:02 ready → published (merged): on the trunk its modules follow: merged there, and its walk opened — or nothing for a walk to move
- 2026-10-09 01:03 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
- 2026-10-09 01:03 delivering → delivered (done): its walk is done: 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/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)
mesh/repo-check: PASS — its merge-check.sh passed
Delivery plan — builds zfs, sent nowhere; no bus step
tier 0: zfs
Run by the build seat on ace as build-1791501499134340514 (builds --log build-1791501499134340514).
--- the gate: zfs
zfs: ok, 15 tool(s), ready: watch by its tool zfs_health every 1m0s; events by unit every 30s; scrub-schedule by unit every 30s
long-running resources without health: 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 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-08T23:13:48Z 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 2026-10-08T23:13:48Z
modules the change moves:
- zfs (changed)
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): zfs
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.002s
words: 12 retired words, none in the 33924 strings of 601 files
ok git.novox.be/novox/mesh-catalog/checks/paths 0.006s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 70, as kept
testing modules/zfs
ok zfstools/cmd/zfs-tools 1.579s
**Merge check** at `e69506a2`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)
- `mesh/repo-check`: **PASS** — its merge-check.sh passed
**Delivery plan** — builds zfs, sent nowhere; no bus step
- tier 0: zfs
Run by the build seat on ace as `build-1791501499134340514` (`builds --log build-1791501499134340514`).
```
--- the gate: zfs
zfs: ok, 15 tool(s), ready: watch by its tool zfs_health every 1m0s; events by unit every 30s; scrub-schedule by unit every 30s
long-running resources without health: 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 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-08T23:13:48Z 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 2026-10-08T23:13:48Z
modules the change moves:
- zfs (changed)
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): zfs
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.002s
words: 12 retired words, none in the 33924 strings of 601 files
ok git.novox.be/novox/mesh-catalog/checks/paths 0.006s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 70, as kept
testing modules/zfs
ok zfstools/cmd/zfs-tools 1.579s
```
The second review of #147 found three blockers before the home server.
1. ${setting:} was not filled in a unit's name, so the distribution timers would have reached the
machine as "zfs-scrub-weekly@${setting:scrub-pool}.timer". mesh-controller #171 fills it and refuses
any value that is not a unit-name part. The module's own schedule needs no setting in a name now: the
node-engine's timer for the scheduled process zfs-mesh-scrub fires `zfs-tools scrub-run` every quarter
of an hour, and the run reads the pool, cadence, day and time from the configuration file.
2. The run does what upstream's zfs-scrub@.service does: if a scrub is running it waits for it with
zpool wait -t scrub, otherwise it starts one and waits. It never pauses or stops a scrub, and the unit
has no ExecStop. A resilver defers the scrub. The served slot is kept in /var/lib/zfs-tools.
3. Unassigning left the pool with no schedule. mesh-host #66 (hq ADR 0269, proposed) gives the
distribution timer back running and enabled. The README says not to unassign before it is merged.
Also: the run comes before the distribution timers in the declaration; the day and time are checked
against fixed sets and reach no unit file (an invalid value runs nothing); a due scrub not started an
hour after its time is a warning; the error counts outlive a restart in a small state file.
mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)
mesh/repo-check: PASS — its merge-check.sh passed
Delivery plan — builds zfs, sent nowhere; no bus step
tier 0: zfs
Run by the build seat on shanks as build-1791502819136072970 (builds --log build-1791502819136072970).
--- the gate: zfs
zfs: ok, 15 tool(s), ready: watch by its tool zfs_health every 1m0s; events by unit every 30s
long-running resources without health: 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 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-08T23:38:08Z 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 2026-10-08T23:38:08Z
modules the change moves:
- zfs (changed)
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): zfs
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.002s
words: 12 retired words, none in the 33983 strings of 601 files
ok git.novox.be/novox/mesh-catalog/checks/paths 0.006s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 70, as kept
testing modules/zfs
ok zfstools/cmd/zfs-tools 1.644s
**Merge check** at `bd018278`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)
- `mesh/repo-check`: **PASS** — its merge-check.sh passed
**Delivery plan** — builds zfs, sent nowhere; no bus step
- tier 0: zfs
Run by the build seat on shanks as `build-1791502819136072970` (`builds --log build-1791502819136072970`).
```
--- the gate: zfs
zfs: ok, 15 tool(s), ready: watch by its tool zfs_health every 1m0s; events by unit every 30s
long-running resources without health: 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 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-08T23:38:08Z 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 2026-10-08T23:38:08Z
modules the change moves:
- zfs (changed)
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): zfs
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.002s
words: 12 retired words, none in the 33983 strings of 601 files
ok git.novox.be/novox/mesh-catalog/checks/paths 0.006s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 70, as kept
testing modules/zfs
ok zfstools/cmd/zfs-tools 1.644s
```
Third review of #147. The run writes as root into /var/lib/zfs-scrub (0700, root's), apart from the
operator account's error counts in /var/lib/zfs-tools, and refuses a folder that is a link, open to
others or someone else's, and a file that is a link: a new file opened O_NOFOLLOW|O_EXCL, renamed over.
mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)
mesh/repo-check: PASS — its merge-check.sh passed
Delivery plan — builds zfs, sent nowhere; no bus step
tier 0: zfs
Run by the build seat on g14 as build-1791503759280452863 (builds --log build-1791503759280452863).
--- the gate: zfs
zfs: ok, 15 tool(s), ready: watch by its tool zfs_health every 1m0s; events by unit every 30s
long-running resources without health: 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 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-08T23:47:08Z 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 2026-10-08T23:47:08Z
modules the change moves:
- zfs (changed)
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): zfs
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.003s
words: 12 retired words, none in the 33988 strings of 601 files
ok git.novox.be/novox/mesh-catalog/checks/paths 0.006s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 70, as kept
testing modules/zfs
ok zfstools/cmd/zfs-tools 1.623s
**Merge check** at `a9acd3dd`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)
- `mesh/repo-check`: **PASS** — its merge-check.sh passed
**Delivery plan** — builds zfs, sent nowhere; no bus step
- tier 0: zfs
Run by the build seat on g14 as `build-1791503759280452863` (`builds --log build-1791503759280452863`).
```
--- the gate: zfs
zfs: ok, 15 tool(s), ready: watch by its tool zfs_health every 1m0s; events by unit every 30s
long-running resources without health: 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 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-08T23:47:08Z 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 2026-10-08T23:47:08Z
modules the change moves:
- zfs (changed)
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): zfs
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.003s
words: 12 retired words, none in the 33988 strings of 601 files
ok git.novox.be/novox/mesh-catalog/checks/paths 0.006s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 70, as kept
testing modules/zfs
ok zfstools/cmd/zfs-tools 1.623s
```
mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)
mesh/repo-check: PASS — its merge-check.sh passed
Delivery plan — builds zfs, sent nowhere; no bus step
tier 0: zfs
Run by the build seat on g14 as build-1791507260556361235 (builds --log build-1791507260556361235).
--- the gate: zfs
zfs: ok, 15 tool(s), ready: watch by its tool zfs_health every 1m0s; events by unit every 30s; WARNING: scrub-config ask(s) for a setting and do(es) not say whether it is trusted, so it counts as trusted: set at the terminal alone; say "trusted" false where nothing trusts it (novox/hq issue 339)
long-running resources without health: 0
files asking for a setting without saying whether it is trusted: 1
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 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-09T00:53:55Z 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 2026-10-09T00:53:55Z
modules the change moves:
- zfs (changed)
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): zfs
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.003s
words: 12 retired words, none in the 33988 strings of 601 files
ok git.novox.be/novox/mesh-catalog/checks/paths 0.005s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 70, as kept
testing modules/zfs
ok zfstools/cmd/zfs-tools 1.668s
**Merge check** at `a9acd3dd`
- `mesh/merge-gate`: **PASS** — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)
- `mesh/repo-check`: **PASS** — its merge-check.sh passed
**Delivery plan** — builds zfs, sent nowhere; no bus step
- tier 0: zfs
Run by the build seat on g14 as `build-1791507260556361235` (`builds --log build-1791507260556361235`).
```
--- the gate: zfs
zfs: ok, 15 tool(s), ready: watch by its tool zfs_health every 1m0s; events by unit every 30s; WARNING: scrub-config ask(s) for a setting and do(es) not say whether it is trusted, so it counts as trusted: set at the terminal alone; say "trusted" false where nothing trusts it (novox/hq issue 339)
long-running resources without health: 0
files asking for a setting without saying whether it is trusted: 1
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 1 module(s) changed against 4 machine(s), as the snapshot of 2026-10-09T00:53:55Z 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 2026-10-09T00:53:55Z
modules the change moves:
- zfs (changed)
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): zfs
--- its merge-check.sh (go toolchain)
ok git.novox.be/novox/mesh-catalog/checks/words 0.003s
words: 12 retired words, none in the 33988 strings of 601 files
ok git.novox.be/novox/mesh-catalog/checks/paths 0.005s
paths: no module names another's state on its machine (3 known, listed to go)
long-running resources without health: 70, as kept
testing modules/zfs
ok zfstools/cmd/zfs-tools 1.668s
```
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
This pull request makes the fixes from both safety reviews of the
zfsmodule (merged as90516fc), and adds the operator's requirement that the module owns the pool's scrub. Do not merge before mesh-controller #171. Do not assign before #171 is live. Do not unassign before mesh-host #66 (hq ADR 0269, proposed: hq #238) is merged.The scrub schedule
Settings:
scrub-pool: the operator's value, with no default.scrub-cadence:weeklyormonthly, defaultweekly.scrub-day:MontoSun, defaultMon.scrub-time:HH:MM, default00:00.The defaults match the distribution's weekly timer on the home server today. The operator chose monthly for ace.
The scheduled run. The node-engine runs the scheduled process
zfs-mesh-scrub(zfs-tools scrub-run, as root) from its ownzfs-mesh-scrub.timerevery 15 minutes. The run finds the slot (the last scheduled time not after now). If that slot is not served yet:zpool wait -t scrub -- <pool>;zpool scrub -- <pool>, records the slot as served, and waits;It never pauses or stops a scrub, and the unit has no ExecStop. The served slot is kept in
/var/lib/zfs-tools/scrub-served.json.No setting value reaches a unit file. The day, time and cadence go only into
/etc/zfs-tools/scrub.conf. The run checks them against fixed sets and refuses anything else (*:*andMon..Friamong them) without running anything. The pool name is the only setting in a unit name: the controller fills it into the distribution timers' names and refuses anything but[A-Za-z0-9:_.-](mesh-controller #171).Exactly one schedule. The distribution's
zfs-scrub-weekly@<pool>.timerandzfs-scrub-monthly@<pool>.timerare held stopped and disabled. They come after the scheduled run in the declaration, so a run that cannot be installed stops the apply before today's timer is touched. Unassigning gives them back running and enabled, once mesh-host #66 is merged.Tools.
zfs_scrubshows the schedule, the next scheduled scrub, the timer's state, any resilver's progress and time to go, and the last scrub and resilver.zfs_scrub_start,zfs_scrub_pauseandzfs_scrub_stoptake{pool, why}and runsudo -n zpool scrub [-p|-s] -- <pool>. They are recorded asscrub.requested, and are candidates for the approval mark (research 039).Health warnings:
A resilver that finished with errors is unhealthy.
Composed for ace by the controller from #171, with the real manifest,
scrub-pool=storageandscrub-cadence=monthly:zfs.scrub-run: processzfs-mesh-scrub, schedule*/15 * * * *(unitszfs-mesh-scrub.timerand.service)zfs.distribution-weekly:zfs-scrub-weekly@storage.timer, stopped and disabledzfs.distribution-monthly:zfs-scrub-monthly@storage.timer, stopped and disabledzfs.scrub-config:pool=storage,cadence=monthly,day=Mon,time=00:00The next scrub would be 2026-11-02 00:00.
The first review's fixes
zfs_snapshot: norecursive(passing it is refused). Every answer gives the dataset's free space and the pool's capacity, and says the snapshot holds every block deleted or overwritten until it is destroyed by hand. A timed-out command is answered fromzfs list, or as "outcome unknown; check zfs_snapshots".Timeouts: a command that runs out of time is ended with its whole process group (SIGTERM, which sudo passes on, then SIGKILL after 5 s).
The watch runs as the operator account.
The event follower:
/run/zfs-eventskeeps it from writing an event twice;The home server's journal is on disk, so the journaled events survive a reboot.
Health: error counts from before are a warning; counts that rise are unhealthy for 15 minutes. The counts are kept in
/var/lib/zfs-tools/errors.jsonacross restarts.The guard test catches every way to run a command (
ask,AsRoot,Escalated,.Run(,.Start(,.Output(,exec.Command(Context)(,os.StartProcess,syscall.Exec). It holds each argument list to its program and verb. The scheduled run may only usescrub(start) andwait.Fixtures and parser follow OpenZFS 2.4.4's source: human scan sizes under
-p, no iostat section lines, error scrubs, per-vdev rebuilds, paused scrubs.Checked
go test -racepasses in UTC, UTC+14, UTC−12 and Europe/Brussels.merge-check.shwith the controller from #171: zfs: ok, 15 tools, health checks declared, long-running resources without health 70 as kept.bd01827,mesh/merge-gateandmesh/repo-checkpass.Delivery
novox/mesh-catalog@a9acd3dd289b— delivered since 2026-10-09T01:03:49ZDelivery plan — builds zfs, sent nowhere; no bus step
Machines
Transitions
The commit's note under
refs/notes/mesh-plankeeps every transition:git log --notes=mesh-plan.Merge check at
e69506a2mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)mesh/repo-check: PASS — its merge-check.sh passedDelivery plan — builds zfs, sent nowhere; no bus step
Run by the build seat on ace as
build-1791501499134340514(builds --log build-1791501499134340514).Merge check at
bd018278mesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)mesh/repo-check: PASS — its merge-check.sh passedDelivery plan — builds zfs, sent nowhere; no bus step
Run by the build seat on shanks as
build-1791502819136072970(builds --log build-1791502819136072970).Merge check at
a9acd3ddmesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)mesh/repo-check: PASS — its merge-check.sh passedDelivery plan — builds zfs, sent nowhere; no bus step
Run by the build seat on g14 as
build-1791503759280452863(builds --log build-1791503759280452863).Merge check at
a9acd3ddmesh/merge-gate: PASS — every machine composes with the change as it did without (4 of 4 compose) (modules: zfs)mesh/repo-check: PASS — its merge-check.sh passedDelivery plan — builds zfs, sent nowhere; no bus step
Run by the build seat on g14 as
build-1791507260556361235(builds --log build-1791507260556361235).