Commit Graph
18 Commits
Author SHA1 Message Date
jschoubben f4ecc6fa64 Pin every mesh-delivery module to mesh-sdk feat/judge-once a20bdf7, the judge out and package versions in
Issue 1585: builder-1585b's package versions folded onto the judge once; one mesh-sdk head for the group.

Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
2026-11-01 02:42:05 +01:00
jschoubben 26852f4a9d Raise a refused manifest as the watchman's finding, so the operator hears it once (ADR 0578 §1)
Issue 1585, S10: the watchman watches the indexer's modules and finds manifest-refused for a module whose value
carries a refusal, ending it once accepted. Its mesh-sdk pin moves to the group head, so it takes health.Finding's
Ref (issue 1952 renamed Token).

Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
2026-11-01 02:09:50 +01:00
jschoubben 8a658a4290 Watchman: an unhealthy finding read back at the start is kept as it was (issue 1952)
mesh/delivery merged: source.main moved to novox/mesh-delivery@91128d4f0af6
mesh/check checking 8a658a4290c9 on 91128d4f0af6
At a restart a standing finding is observed again by the replayed failed have: kept, not confirmed again, and its urgent
signal runs on from the finding's Since; an urgent one stays urgent with no signal restarted.
Quick: skip the confirm when a finding stands. Thorough, chosen: the same, plus urgency read from the finding itself.

Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
2026-10-31 22:01:59 +01:00
jschoubben fb8d2386b9 Watchman: a module is unhealthy when its have still says failed a minute after it first did (issue 1952)
A have that turned failed once and stayed unchanged was written once, so the old count of two writes never fired and the
finding was never written (fail2ban on novox). Confirm by time, with a signal, also at a restart.
Quick: raise the count to one write (loses the confirmation). Thorough, chosen: confirm by a time bound, as ADR 0564 s4's
signals are keys that expire, so no write count is needed.

Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
2026-10-31 21:59:46 +01:00
jschoubben 255f8b2aca Close a stuck key whose watched key aged out or was deleted unheard, so its condition clears (issue 774)
A key that ages out of its bucket (the build seats' asks pass the stream's 168 h) leaves no marker a watch hears,
and one deleted while the watchman was down is not heard either: its stuck key stood unmoved for ever, and 26
stuck-key conditions of asks from 2026-10-21 were open on 10-30.

Quick fix: mark stuck keys older than the stream's maximum age moved.
Thorough fix, chosen because it covers every bucket the watchman mirrors and any way a key goes: at the start's
end, a stuck key not moved whose bucket was read whole at this start (each build seat's steps, the assigner's
wants) and whose key was not read standing is written moved, which clears its finding, and the closing is said
in the log. A bucket that could not be read closes nothing. A key that ages out while the watchman runs is
closed at its next start: a KV age-out emits no event, and no timer is added.
2026-10-30 18:25:56 +01:00
jschoubben ee2f1efe78 Watchman: hold a failing network part at the machines it points at, as the controller does, so a machine others cannot reach is said once there and not as each reporter's own (issue 774)
Pins mesh-sdk 6eae076.
2026-10-29 23:06:19 +01:00
jschoubben 8a91871de6 Watchman: raise machine-network, network-rewritten, network-unreachable, machine-units, relogin-needed and directory-used-as-found from each node's own health report, under the controller's keys, so these alerts outlive the controller's lease (issue 774)
Pins mesh-sdk 94d0de5.
2026-10-29 23:01:03 +01:00
jschoubben 78f03b7422 Watchman: say needs-operator from mesh-provisioner's provisions in waits, so the row the haves could not carry has its facts on the bus (issue 1954 D2)
Pins mesh-sdk 4a96793.
2026-10-29 22:37:39 +01:00
jschoubben 15a15cbda5 Watchman: observe S1, S9 and S21 as the controller does and stop the needs-operator row, so the keeper's compare can empty with no condition lost (issue 1952 WM2, WM3, WM5)
S1 is urgent after half an hour for a control node (the node running mesh-controller). S9 also hears the bus module's
slow consumers and closed connections, each closure counted once. S21 is one finding per heard machine from its failed
haves and its provisioner clients' reports (PROV_APPLIED). DEAD_LETTERS is counted again at a letter's age and every
five minutes while letters are held, on the watchman's own pulse. needs-operator waits for L6: its inputs are not on the
bus. Pins mesh-sdk b6ef01d.
2026-10-29 21:59:46 +01:00
jschoubben da787ae360 Watchman: hold mesh-watchdogs and write its watchdogs as findings, so health can leave the controller (issue 1952 WM1-WM5)
The signals rows S1, S8, S9, S10, S11, S21 and the haves' unhealthy and needs-operator are pulses whose expiry is a finding;
stuck keys and presence are its own findings; it keeps the dead letters from the lease write on its own consumer, lists and
shows them, and refuses deliver and drop until the hand-act log's home (issue 1953). Tools read every state it keeps and
start resync. Pins mesh-sdk 2a04bf0.
2026-10-29 21:44:18 +01:00
jschoubben 806c8bf0e4 Pin the assigner, watchman and view to mesh-sdk's release 6 head (issue 1943 S3) and give their test wants a composition digest 2026-10-24 16:32:09 +02:00
jschoubben 9d2e9dc6c9 The watchman mirrors each want of the assigner as a pulse owed by the node's agent until the have names its fingerprint (issue 1906 D6) 2026-10-24 16:22:02 +02:00
jschoubben c7bddbbeab Bound a running build step's pulse by the step's own bound from its start, since nothing rewrites the key while it runs (issue 1908) 2026-10-23 16:58:26 +02:00
jschoubben d32ec6b7f8 Say what the builder, the indexer and the watchman do since R4g in their comments: no build-agent is asked, no build record moves a fingerprint, no checks-running bucket is mirrored (issue 1887, review round 2 item 6) 2026-10-23 16:47:36 +02:00
jschoubben e927469460 Pulse a build ask only from when a node takes it, and end the pulse with its step, so a waiting ask is a backlog and not a stuck row (issue 1908) 2026-10-23 16:34:39 +02:00
jschoubben a8ba5c2fd5 Mirror a waiting build ask as a pulse until a node writes its step, and drop the read of the controller's checks-running (issue 1887 A8b) 2026-10-23 16:12:43 +02:00
jschoubben e5fd3fcd7a The watchman mirrors no builds in release 2; a check's pulse is renewed at every heartbeat of its key (issue 1135, review round 1) 2026-10-18 21:05:50 +02:00
jschoubben bdb6fe7787 The watchman's logic: presence from the bus's events, pulses with a time to live, stuck keys on expiry (issue 1135) 2026-10-18 20:58:34 +02:00