| 21:01 |
<ryankemper> |
[WDQS] Alright, these are re-entering a failed state soon enough that we will need to identify the offender if we want to restore proper service. We could put some temporary hack to restart every few minutes so we at least maintain some uptime, but root cause is the usual 'we need a requestctl rule to block whoever's killing us' scenario |
[production] |