Shared computer at its memory cap: docker exec times out, every bot shows "team computer is being changed", restart/kill hang under gVisor

Aperta
#24 2 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

Nessuno ha ancora preso questa issue.

  • #37 di @warhammered — chiusa senza merge

Valutazione

Difficoltà
5/5
Tempo stimato
Più di una settimana
Idoneità per principianti
35/100
Tipo di issue
Bug
Chiarezza
Abbastanza chiara
Stato di attività
Attiva
Stack tecnologico
docker, linux, typescript

Direzione di ricerca

The report describes memory limits, browser allocation, container recovery, lifecycle locking, and restart behavior, but names no source files, tests, or entry points. Start by locating the shared computer lifecycle and container management code, then trace how browser requests and Docker restarts are handled. Done means the reported failure modes have defined fixes and regression coverage; the issue does not specify acceptance criteria.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

Version

0.2.15

Platform

Own server (Windows 11 + WSL2 Ubuntu 24.04, 15.8 GB RAM, WSL memory=10GB at the time)

What happened

On 17 Nov about 16:50 WIB the shared computer container (openmausbot-vps-computerf3d9-..., runtime runsc) reached 7.18 of its 7.2 GB limit with 355 processes, because every bot opened its own browser in it. Then:

  • docker exec timed out, mcp__agents__vm_exec calls failed after up to 338 s (94 calls, median 1.9 s, p90 60 s, 81% ok);
  • every bot in every group showed "The team computer is being changed" and stopped;
  • docker restart and docker kill both hung. Only kill -9 of the runsc-sandbox process plus docker start recovered it.

Earlier the same day (15:17 WIB) the in-memory remoteComputerLifecycleBusyOwners flag for the same computer was never released although the container was up, and mountTeamVps kept throwing until duskbot was restarted.

Suggested fix
  1. Watch the container's memory and refuse or queue a new browser screen when it is above about 85%, with a message saying why.
  2. Default bots that never use a browser to no team computer (we set computer:"off" on 10 of 20 bots by hand).
  3. When docker stop/kill does not return in 20 s on runsc, kill the sandbox process and restart the container automatically.
  4. Release the lifecycle busy flag in a finally, and expire it after a timeout.
  5. Keep a memory limit set with docker update when the container is restarted, instead of resetting it.
Lingua principale
TypeScript
Stelle
277
Fork
27
Merge medio
2h 47m
PR unite (30g)
1

Preparare l'ambiente

Questo progetto non fornisce container di sviluppo, Dockerfile né guida per i contributori, quindi l'ambiente è a tuo carico: parti dal suo README e consulta la nostra guida al primo contributo per i passaggi generali.

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di DuskBotAI/duskbot

Tutte le issue di DuskBotAI/duskbot

Issue simili

Altre issue su TypeScript

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.