next bear
Specs: 5 Core AMD EPYC (Genoa), 60Gb DDR5, 250Gb SSD, Ubuntu 22.04 running Docker server (https://github.com/indifferentbroccoli/windrose-server-docker).
Random Container Restart/Game Crash for all players, sometimes when boarding, sometimes in ship combat, sometimes just walking around. Seems to be an Async Queue issue in RockDB. Attached sample logs here and will upload incident bundle with timestamps/logs/more info if possible.
From logs, repeated crash sequences are triggered by R5Check failures in Rocks DAL async processing:
Condition: 'Status.ok()'
Message: RecordId '<ID>'. Error 'NotFound: '
Where: R5BLRocksAsyncQueue::DoJob::<lambda_8>::operator () at R5BLRocksAsyncQueue.cpp:341
Each incident is followed by FR5CheckDetails::PrintCallstackToLog / crash stack trace and then a fresh startup in the next rotated log.
Recent recurrence window was April 24-25, 2026, with multiple unique failing RecordIds, indicating recurring data/state inconsistency rather than one-off corruption.
No clear OOM-kill signature was found in app logs.
I’ve attached an incident bundle with:
timeline and parsed crash events,
failing RecordIds,
related Sentry issue IDs/URLs,
per-incident raw log snippets and source logs.