audit 2026-11-15: capture stall 38h46m (freeze total, counts unchanged); cron_incidents 1->3 unclosed (signature embeds elapsed-hours float); monitor's 4.64h decoded as newest-conversational-message age; watchdog lost an 11:45 slot to a 600s deps-guard timeout; ledger 168 rows 165/3 (11-14 settled 167/0)

This commit is contained in:
Tony0410
2026-11-16 00:32:12 +08:00
parent 402d8546e8
commit 9f80926dda
6 changed files with 180 additions and 22 deletions
+9 -4
View File
@@ -3,9 +3,9 @@ title: Mnemosyne
type: system
status: active
created: 2026-07-22
updated: 2026-11-15
verified_on: 2026-11-15
review_after: 2026-11-13
updated: 2026-11-16
verified_on: 2026-11-16
review_after: 2026-11-18
confidence: high
tags: [memory, mnemosyne, backend]
sources: [live-probe, mnemosyne-db, config, ops-profile-config, live-secrets-scan, live-state-db, live-health-monitor-log]
@@ -50,7 +50,7 @@ Primary agent memory store for Hermes.
## Known Issues
**⚠ OPEN — capture stalled at 2026-11-14 09:32:59 AWST and the monitor took 14 h 31 m to escalate (audit 2026-11-15 00:4x).** `working_memory` and `episodic_memory` stop in the same minute (last 8 working writes all 01:01–01:32 UTC); ~15 h of desktop / Telegram / cron turns are **not** in memory, and the 23:30 sleep cycle consolidated **0 items**. The health monitor `ea79a2107e36` exited 1 at **2026-11-15 00:03:44** on `STALE MEMORY PIPELINE` (incident `ea79a2_601037516abd`, `state: alerted`, unclosed; its `failure_type: auth` is the scheduler's classifier, not evidence of a credential fault). **Correction, re-derived 2026-11-15 01:2x: the 12:00:52 slot saw nothing** — it measured `capture delay 29.4 min` (healthy, exit 0, delivery `suppressed`), and the `-589.5 min` ℹ️ line belongs to the **00:03:35 run itself** (monitor log: header line 957 = 12:01:52 run, header line 1030 = 00:03:35 run, `-589.5` at line 1046, ❌ at 1047). `mnemosyne_freshness_gate.py` is **magnitude-based** (`delay_min < -CAPTURE_LAG_MIN`, default **120** min) but **activity-aware**: it can only fire once a *conversational* message postdates the last capture. The last captured message was the 09:03 AWST `ios` turn (captured 09:32:59, +29.4 min) and the first to postdate the stall was the **19:22 AWST desktop** turn — after the 12:00 slot. **The quiet was the `0 */12` cadence, not a sign-vs-magnitude bug.** **What did go unread was volume: 29 consecutive 0-item consolidation cycles from 10:04:42 AWST**, printed as `(0 items)` on every run yet tested only for age (`> 36 h`) — the candidate fix is a zero-item-streak threshold. ⚠ Related trap, verified: `consolidation_log.created_at` mixes formats (recent rows `+00:00`, older rows naive **AWST**), so treating the column as one axis misplaces events by 8 h. The known stringified-`config.yaml` killer is **ruled out** (`config.yaml` mtime 2026-11-13 18:03; full YAML walk shows no `[`/`{` scalars). Root cause **not established** — no error lines; the writes simply stop. Alert path verified working (delivery `for_failure: 1`, `delivered` 00:03:39). → [[incidents/2026-11-14-mnemosyne-capture-stall]]
**⚠ OPEN — capture stalled at 2026-11-14 09:32:59 AWST and the monitor took 14 h 31 m to escalate (audit 2026-11-15 00:4x).** `working_memory` and `episodic_memory` stop in the same minute (last 8 working writes all 01:01–01:32 UTC); ~15 h of desktop / Telegram / cron turns are **not** in memory, and the 23:30 sleep cycle consolidated **0 items**. The health monitor `ea79a2107e36` exited 1 at **2026-11-15 00:03:44** on `STALE MEMORY PIPELINE` (incident `ea79a2_601037516abd`, `state: alerted`, unclosed; its `failure_type: auth` is the scheduler's classifier, not evidence of a credential fault). **Correction, re-derived 2026-11-15 01:2x: the 12:00:52 slot saw nothing** — it measured `capture delay 29.4 min` (healthy, exit 0, delivery `suppressed`), and the `-589.5 min` ℹ️ line belongs to the **00:03:35 run itself** (monitor log: header line 957 = 12:01:52 run, header line 1030 = 00:03:35 run, `-589.5` at line 1046, ❌ at 1047). `mnemosyne_freshness_gate.py` is **magnitude-based** (`delay_min < -CAPTURE_LAG_MIN`, default **120** min) but **activity-aware**: it can only fire once a *conversational* message postdates the last capture. The last captured message was the 09:03 AWST `ios` turn (captured 09:32:59, +29.4 min) and the first to postdate the stall was the **19:22 AWST desktop** turn — after the 12:00 slot. **The quiet was the `0 */12` cadence, not a sign-vs-magnitude bug.** **What did go unread was volume: 29 consecutive 0-item consolidation cycles from 10:04:42 AWST**, printed as `(0 items)` on every run yet tested only for age (`> 36 h`) — the candidate fix is a zero-item-streak threshold. ⚠ Related trap, verified: `consolidation_log.created_at` mixes formats (recent rows `+00:00`, older rows naive **AWST**), so treating the column as one axis misplaces events by 8 h. The known stringified-`config.yaml` killer is **ruled out** (`config.yaml` mtime 2026-11-13 18:03; full YAML walk shows no `[`/`{` scalars). Root cause **not established** — no error lines; the writes simply stop. Alert path verified working (delivery `for_failure: 1`, `delivered` 00:03:39). **Update 2026-11-16 00:18 (next audit): the freeze is total, not degraded** — `working_memory` **1130** / `episodic_memory` **1249**, *identical at every read of the stall*, so **0 writes in 38 h 46 m**; newest rows are still the `[MODEL_REFRESH_PROPOSAL]` maintenance writes at `09:32:58`/`09:32:59` and the newest *user-turn* capture is `2026-11-14 09:04:19`. The monitor failed twice more (`12:03:14`, `2026-11-16 00:03:15`) → `cron_incidents` **1 → 3 unclosed** (141 rows total), each a single observation (`first_seen_at == last_seen_at`) because **the signature embeds the elapsed-hours float** (`4.64157453` / `4.634034074444445` / `4.643481075555555`), which is never equal twice — one durable fault, a fresh row every slot, nothing ever escalates. ⚠ **And the `≈4.64 h` in the alert is not a stall duration:** `mnemosyne_freshness_gate.py`'s `newest_user_message()` reports the age of the newest message whose session source is not `cron`/`subagent` — the **Pulse/`desktop`** turns, ~6 h cadence — so **the alarm has understated severity on every fire** (mapping table on the incident page). The consolidator's `summary_preview` now reads `No-op consolidation cycle — timestamp updated for watchdog`: it **bumps its own timestamp to satisfy an age test**, so a zero-item streak is invisible to an age-only gate; streak now **~74 cycles** from `2026-11-14T02:04:42+00:00`. **Proposed, none applied:** a zero-item-streak threshold, and normalising the value out of the incident signature. → [[incidents/2026-11-14-mnemosyne-capture-stall]]
**⚠ Security - live `.env` values are stored verbatim in this store's own tables (live read 2026-11-06 00:2x-00:5x AWST).** **17 rows** carry three live credentials - `TELEGRAM_BOT_TOKEN` (4 rows), `MATTERMOST_TOKEN` (6) and `HERMES_CUSTOM_OMNIROUTE_20128_API_KEY` (7) - across `memories.content`, `working_memory.content`, `episodic_memory.content`, `gists.text`, `memoria_facts.context_snippet` and the `fts_episodes` / `fts_working` mirrors (per-key breakdown on the incident page). The 0-byte `episodic_memory.db` and `memoria_facts.db` stubs beside `mnemosyne.db` hold nothing - those names are *tables inside* the main DB. Capture stores verbatim session text (the 1.7 GB session store `state.db` itself carries all four values), so any session in which a key was printed becomes recall material. **Recall is the exposure path:** these rows can be pulled into any prompt, cron delivery or subagent transcript, and Mnemosyne's own rotation snapshots (seven uncompressed `.db` files under `backups/`) reproduce the whole DB. Not remediated. -> [[incidents/2026-11-05-secrets-at-rest-in-mnemosyne-and-cron-prompts]]. Capture-time redaction of `.env` values is the durable fix; **rotation is the only remedy that also covers the existing copies**.
@@ -102,6 +102,10 @@ Primary agent memory store for Hermes.
- Mnemosyne DB Backup weekly Sunday 03:00 — script: `mnemosyne-backup.sh`, mode: no-agent
- Mnemosyne Health Monitor every 12h — **script**: `mnemosyne-health-monitor.sh`, mode: **no-agent** since
the 2026-11-08 conversion (job `ea79a2107e36`), reports only on exception.
⚠ **2026-11-15 → 11-16: failing on a three-slot streak** — `last_status: error`, `failure_streak`
**3** (failed `00:00:53.928`, `12:01:24.719`, `2026-11-16 00:01:33.030`), the fleet's **only**
non-clean row. Each failure raises a **new** unclosed `cron_incidents` row for the same condition
(signature fragmentation — see the capture-stall incident page).
⚠ **The no-agent conversion was verified end-to-end on 2026-11-08:** the 18:00 slot (agent mode) ran
**12:00:51 → 18:00:42 = 5 h 59 m 51 s**, and the converted job's first slot ran **2026-11-09 00:00:48 →
00:02:02 = 73 s**, exit 0, **silent** — `output/ea79a2107e36/2026-11-09_00-02-00.md` is 159 bytes against
@@ -119,6 +123,7 @@ Primary agent memory store for Hermes.
the 16:37:21 catch-up. **Check `finished_at` before believing a long-running slot failed.**
- Mnemosyne Embedding Coverage every 6h — job `b63ed5817212`, alerts to Telegram topic 102 only when working-tier coverage drops below 95%
- Mnemosyne Watchdog every 15m — script: `mnemosyne-watchdog.py`, mode: no-agent
⚠ **2026-11-15 11:31:20 slot failed after 17 m 49.8 s** — `runtime deps guard could not run: Command '['bash', '/home/hermes/.hermes/scripts/hermes-runtime-deps-guard.sh']' timed out after 600 seconds`. The `11:45` slot was **skipped** and `12:01:25` completed, so it self-recovered with one slot lost. **New signature for this job** — the 2026-08-19 / 10-16 occurrences had the guard quoting *another* job's error. 1 failure in 97 rows across 11-15 → 11-16.
- ⚠ **Blast-radius note (2026-10-16).** Both monitors above (`ea79a2107e36` Health every 12 h,
`b63ed5817212` Embedding Coverage every 6 h) are pinned to **`custom:omni-101`** — the same node the
memory system's own reporting depends on. During the **2026-10-14 → 10-15 `omniroute-1` outage both
+4 -3
View File
@@ -3,9 +3,9 @@ title: Scheduled Tasks
type: system
status: active
created: 2026-07-22
updated: 2026-11-15
verified_on: 2026-11-15
review_after: 2026-11-15
updated: 2026-11-16
verified_on: 2026-11-16
review_after: 2026-11-17
confidence: high
tags: [cron, scheduled, automation]
sources: [live-cron-list, live-jobs-json, cron-executions-db, live-executions-db, live-deliveries-db, live-sessions-db]
@@ -14,6 +14,7 @@ sources: [live-cron-list, live-jobs-json, cron-executions-db, live-executions-db
# Scheduled Tasks
## Count
- **Current as of 2026-11-15 (window 2026-11-14 23:59 -> 2026-11-15 23:59; nightly audit).** **168 rows — 165 `completed` / 3 `failed`**, deliveries **9 / 9 `delivered`** (**3** `for_failure`). The three failures, in order: `ea79a2107e36` (Mnemosyne Health Monitor) `00:00:53.928 -> 00:03:44.604` exit 1; `96eb95ce12f2` (Mnemosyne Watchdog) `11:31:20.699 -> 11:49:10.534610` — **17 m 49.8 s** — exit 1 with `runtime deps guard could not run: Command '['bash', '/home/hermes/.hermes/scripts/hermes-runtime-deps-guard.sh']' timed out after 600 seconds` (**a new signature for that job**; the `11:45` slot was **skipped** and `12:01:25.120952` completed, so it self-recovered with one slot lost); `ea79a2107e36` `12:01:24.719 -> 12:03:20.430` exit 1. **2026-11-14 settled 167 / 167 `completed` / 0 `failed`** (0 `running` — that audit finished, so nothing was silently in flight). Fleet **29 jobs (26 enabled / 3 disabled)** — a **tenth consecutive window at the same count**; the only non-clean `last_status` remains `ea79a2107e36` (`error`, `failure_streak` **3**). `cron_incidents` **1 -> 3 unclosed** (**141** rows) — all three this job, all `failure_type: auth`, each `first_seen_at == last_seen_at`; the signature embeds the elapsed-hours float so the same fault fragments a new row per slot (full analysis on [[incidents/2026-11-14-mnemosyne-capture-stall]]). Sessions started 11-15: **9, all `source: cron`** — no interactive session. 2026-11-16 so far (live 00:2x): 7 rows — 5 `completed` / 1 `failed` (the monitor's `00:01:33.030` slot) / 1 `running` (this audit).
- **Current as of 2026-11-14 (window 2026-11-13 23:59 -> 2026-11-14 23:59; nightly audit).** **A fourth
consecutive zero-`failed` day for the fleet — then the memory monitor broke the clean streak at 00:03.**
Fleet **29 (26 enabled / 3 disabled)**, a **ninth consecutive window at the same count** (nothing added,