Working memory 2026-10-05
Source: Mnemosyne backup 2026-11-06 (hermes-data-mnemosyne.db). Part of Mnemosyne Archive.
2 entries.
d81f1924618ea8b0
> [USER] [IMPORTANT: You are running as a scheduled cron job. DELIVERY: Your final response will be automatically delivered to the user — do NOT use send_message or try to deliver the output yourself. Just produce your report/output as your final response and the system handles the rest. SILENT: If there is genuinely nothing new to report, respond with exactly "[SILENT]" (nothing else) to suppress delivery. Never combine [SILENT] with content — either report your findings normally, or say [SILENT] and no
id: d81f1924618ea8b0 · source: conversation · timestamp: 2026-10-05T09:15:37.526783 · session_id: hermes_cron_11a031b2a485_20261005_090057 · importance: 0.5 · veracity: unknown · created_at: 2026-10-05 09:15:37 · memory_type: instruction · consolidated_at: 2026-10-10T19:30:56.137436 · recall_count: 13 · last_recalled: 2026-10-18T09:01:14.297966 · scope: session · channel_id: hermes_cron_11a031b2a485_20261005_090057 · trust_tier: STATED · event_date_precision: unknown
db976e20f85acb28
> Improved /usr/local/bin/disk-watchdog (cron every 15min, fires at >=85% root disk usage). Old version: unconditionally vacuumed journal + deleted ALL rotated logs system-wide (.1/.gz/*.old) + pruned docker at 85%, paused qbittorrent at 96%, no self-logging, no resume logic. This destructive log-deletion is why Sep 28 crawl4ai incident logs and all nginx history were unrecoverable during a debugging session on 2026-10-05.
>
> New version (deployed 2026-10-05, old one backed up as /usr/local/bin/disk-watchdog.bak_<timestamp>):
> - Tiered thresholds: 85% = journal vacuum + docker prune only (rotated logs now kept). 93% = also deletes rotated logs (held back from the old 85% trigger). 96% = pauses qbittorrent + logs top-10 disk consumers via du. <90% = auto-resumes qbittorrent if currently paused (checks live docker state, not a statefile).
> - Self-logs every action to /var/log/disk-watchdog.log (excluded from its own deletion glob) with usage%, action taken, and MB reclaimed, so future debugging isn't blind.
> - Root cause of the 96% disk usage found same day: /opt/media-stack/data (qbittorrent/sonarr/radarr/jellyfin media library) = 185GB of 222GB used. A cleanup script exists at /opt/scripts/cleanup_imported.py (deployed 2026-10-02) meant to de-prioritize/prune completed torrents after import but apparently isn't keeping pace. Media library itself was NOT touched/deleted — that's the user's content, left for them to decide on.
id: db976e20f85acb28 · source: task · timestamp: 2026-10-05T21:20:30.661077 · session_id: mcp_default · importance: 0.8 · metadata_json: {"_dates": ["2026-10-05", "2026-10-05", "2026-10-02"]} · veracity: tool · created_at: 2026-10-05 21:20:30 · memory_type: relationship · consolidated_at: 2026-10-10T19:30:56.410600 · recall_count: 2 · last_recalled: 2026-10-06T11:28:31.536396 · scope: global · channel_id: default · trust_tier: STATED · event_date: 2026-10-05 · event_date_precision: day · temporal_tags: ["2026-10-05", "week-41-2026", "monday"]