stocktake(2026-11-15): skills-index regenerated (299/67); storm incident re-verified (all 12 rows resolved, evidence-window note); conduit memory-timeout fix LOST in plugin 0.168.x restructure — page → degraded with regression section; 0 stale after fixes

This commit is contained in:
Tony0410
2026-11-15 03:25:18 +08:00
parent 646be98c31
commit 402d8546e8
6 changed files with 233 additions and 23 deletions
+31 -1
View File
@@ -1751,4 +1751,34 @@ write path, and nothing watches it.** Candidate fix: an N-consecutive-zero-item-
14 `script` / 5 `auth`. The job is `no_agent`, so `agent` is impossible — read `auth` as unexplained, not as a
credential fault. Pages corrected: [[incidents/2026-11-14-mnemosyne-capture-stall]] (false section rewritten, `auth` bullet upgraded
from suspicion to evidence), [[systems/scheduled-tasks]], [[systems/mnemosyne]], [[index.md]]. The incident itself
remains **OPEN / `alerted` / unclosed**; the *cause* of the stall is still **not established**.
remains **OPEN / `alerted` / unclosed**; the *cause* of the stall is still **not established**.
## 2026-11-15 03:00 — stocktake (wiki-maintain.sh + drift re-verification)
Script run: skills-index regenerated (299 skills / 67 categories), freshness report
`raw/stocktake/freshness-20261115.txt` — **STALE (1)**:
`incidents/2026-10-09-cron-provider-timeout-storm.md` (22d, updated 2026-10-24). Pages ON the
threshold (21d): `incidents/2026-10-22-conduit-memory-start-timeout.md`,
`reports/communities-wa-social-media-audit-2026.md` (static deliverable, no live claims — left
unbumped). Pages at 20d (not re-probed this run, next candidates):
`runbooks/provider-health.md`, `runbooks/update-hermes-safely.md`,
`systems/current-profile.md`, `systems/memory-backend.md`, `systems/model-providers.md`,
`incidents/2026-10-03-ct450-hermes-update.md`,
`incidents/2026-10-03-mnemosyne-vector-recall-fts-only.md`,
`incidents/2026-10-24-conduit-call-request-refused.md`,
`incidents/2026-10-25-gmail-digest-runaway-and-refire.md`.
**Re-verified live (storm page, 22d):** all 12 `cron_incidents` rows still `resolved`; weekly job
`1b23c042e17d` `last_status: ok` (2026-11-13). The one open fleet-wide incident
(`ea79a2_601037516abd`, first seen 2026-11-15 00:03) is the capture-stall monitor firing, not a
storm row. `executions.db` retention now starts 2026-11-09 — the page's 2026-10-09 execution
counts are no longer DB-re-derivable; noted on the page. Bumped with a dated verification note.
**Regression found (conduit page, 21d):** the Conduit voice memory-timeout fix was **lost in the
plugin's restructure**. `~/.hermes/plugins/conduit_push` is clean at upstream `0.168.2`; the
memory layer moved to `dashboard/api/memory.py` where `MEMORY_TIMEOUT_S = 10.0` is hard-coded
again (no `CONDUIT_MEMORY_TIMEOUT_S` env hook; restructure landed ~2026-10-29 AWST). Running
dashboard (since 2026-11-10) carries it. One real call since (2026-11-14 08:57 AWST) hit a warm
provider, so the cold-start path is untested. Page → `status: degraded` with a dated regression
section; **fix not re-applied** (needs Anthony's call: re-port + commit locally, or upstream).
Pages corrected: [[incidents/2026-10-09-cron-provider-timeout-storm]],
[[incidents/2026-10-22-conduit-memory-start-timeout]], [[index.md]].