memory: clean up active lock files on exit and termination signals

Aperta
  • Qualcosa non va?
#5,485 1 commento 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

@Faris-as ci sta già lavorando.

Dal 6/11/2026.

  • #1 di @Faris-as — aperta

Valutazione

Difficoltà
3/5
Tempo stimato
1-2 giorni
Idoneità per principianti
25/100
Tipo di issue
Bug
Chiarezza
Specificata chiaramente
Stato di attività
Ferma
Stack tecnologico
docker, node.js, typescript
Ambito
backend, devops

Direzione di ricerca

Read src/memory/index.ts, the lock handling in file-lock.test.ts, and src/memory/Dockerfile; compare signal handling with src/everything/transports/stdio.ts and src/everything/Dockerfile. Check the linked prototype branch for existing work. Done means active locks owned by this process are cleaned up on shutdown, locks owned by others are preserved, the Docker stop signal is configured, and the memory tests pass.

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

Descrizione

Which server?

memory (@modelcontextprotocol/server-memory)

Server version

2.1.0

How did you run the server?

From a clone of this repository

Transport

stdio

Protocol era

Modern (2026-07-28)

MCP client and version

Claude Desktop / Autonomous Agent Orchestrator

Operating system and runtime

Linux / Node.js v24+

Server configuration
{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-memory"]
    }
  }
}
Steps to reproduce
  1. Run @modelcontextprotocol/server-memory.
  2. Dispatch a mutation tool call (e.g. create_entities or add_observations) that acquires the lock via withFileLock.
  3. Terminate or stop the server process via SIGINT, SIGTERM, client disconnect, or container stop during the operation.
  4. Inspect the directory holding memory.jsonl.
  5. Attempt a subsequent write from another agent or container sharing the same graph file.
Expected behavior

The server should register signal handlers (SIGINT, SIGTERM) and exit cleanup hooks to remove any active lock file it currently holds, preventing orphaned lock files on abrupt shutdown.

Actual behavior

In Node.js, asynchronous finally blocks in withFileLock do not run when terminated by signals. Because src/memory registers no signal handlers (unlike src/everything/transports/stdio.ts), memory.jsonl.lock remains stranded on disk.
In multi-container or Docker environments where hostname !== os.hostname(), isLockOwnerDead returns false, forcing other agents to stall for up to 30–60 seconds before breaking the stale lock.
Additionally, in src/memory/Dockerfile, Node runs as PID 1 without STOPSIGNAL SIGINT, causing PID 1 to ignore SIGINT by default.

Already prototyped a fix?

Prompt for maintainer workflow:

"In src/memory/index.ts, track active lock files held by this process in an activeLocks map. On process exit and on SIGINT/SIGTERM, synchronously check if the lock file still contains our token, and if so, unlink it to avoid leaving stranded locks on abrupt shutdown. Add STOPSIGNAL SIGINT to src/memory/Dockerfile matching src/everything/Dockerfile. Add unit tests in file-lock.test.ts verifying active lock tracking and that cleanup leaves locks owned by another process intact."

Verification:

Before you submit
  • I searched existing issues and this is not a duplicate.
  • This is not a security vulnerability report.
Lingua principale
TypeScript
Stelle
91.7k
Fork
11.9k
Merge medio
3h 30m
PR unite (30g)
199

Preparare l'ambiente

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 modelcontextprotocol/servers

Tutte le issue di modelcontextprotocol/servers

Issue simili

Altre issue su TypeScript

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.