vitya b0559fdbed wiki(books-vds): disk-full incident 2026-06-04 + Bookva->slovo job scoping
Provider Rusonyx alert (96% used). Reclaimed ~64G: truncated runaway
task-runner json-logs (38G) + removed orphan BuildKit builders/volumes
(26G). Added logrotate copytruncate guard.

Root cause: books-task-runner (slovo) error-looped on Ozon seller Bookva
(50542) whose key was intentionally swapped to slovo's as an access
revocation. Spam came from unscoped legacy seller-iterating agenda jobs.
Fixed: idSeller:2 on 5 jobs (deploy-durable, non-_fromConfig). Verified
6448 errors/min -> 0.

Group-V (channel-driven) confirmed non-issue: Bookva channel is type=ym,
no active Bookva-ozon channel exists.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-04 12:28:30 +03:00
@
2026-05-30 08:35:12 +03:00

admin

Infrastructure agenda — VDS / NAS / recovery-host / OpenWRT + fault-tolerance roadmap.

See .wiki/overview.md for orientation.

Description
Infrastructure agenda � VDS/NAS/recovery-host/OpenWRT + fault-tolerance roadmap
Readme 11 MiB
Languages
PowerShell 77.3%
Shell 22.7%