mcp-memory-service Docker boot loop: container restarts after exit code 0

Question

Running mcp-memory-service with the default docker-compose configuration causes a boot loop: the container initializes ChromaDB, logs "Server started and ready to handle requests", then exits with code 0 and Docker restarts it repeatedly. I changed no environment variables. How do I keep the container running?

Originally asked on GitHub: https://github.com/doobidoo/mcp-memory-service/issues/52

AgentArmorHQ team · Accepted Answer

**Status: Fixed as of 2025-07-11 (no release version mentioned in the thread).** The maintainer implemented the fix and marked the issue resolved.

The root cause, per the maintainer: the MCP server expects an active stdio connection from an MCP client. When run inside Docker without an attached MCP client, the stdio streams close immediately, the server exits gracefully with code 0, and Docker’s restart policy turns that into a boot loop.

Steps from the thread:

1. Use the standalone Compose file the maintainer added (recommended in the thread for testing/development):
   ```bash
   docker-compose -f docker-compose.standalone.yml up
   ```
   This sets `MCP_STANDALONE_MODE=1` and uses a persistent entrypoint script so the server stays up without an MCP client.

2. Or enable standalone mode manually:
   ```bash
   docker run -d -p 8000:8000 \
     -v ./data/chroma_db:/app/chroma_db \
     -v ./data/backups:/app/backups \
     -e MCP_STANDALONE_MODE=1 \
     mcp-memory-service
   ```

3. Verify the container is no longer restarting:
   ```bash
   docker ps
   ```

The maintainer also disabled ChromaDB telemetry, which removes the `Failed to send telemetry event ... capture() takes 1 positional argument but 3 were given` errors from the logs. Standard `docker-compose up` also received improved stdio handling and better signal handling, per the thread. The reporter said they would test the fix later and had not confirmed it at the time of writing, so the fix is based on the maintainer’s closure of the issue.

AgentArmorHQ team · Answer

**What a GitHub user tried on GitHub** (Didn't work)

Ran `docker-compose up` with the default configuration and no environment variable changes; observed repeated initialization, exit code 0, and restart.

Source: https://github.com/doobidoo/mcp-memory-service/issues/52

AgentArmorHQ team · Answer

**What @doobidoo tried on GitHub** (Outcome unclear)

Maintainer's fix: use `docker-compose -f docker-compose.standalone.yml up` or set `MCP_STANDALONE_MODE=1` in a docker run command to keep the server running without an MCP client. Not yet confirmed by the reporter.

Source: https://github.com/doobidoo/mcp-memory-service/issues/52#issuecomment-3060826998

mcp-memory-service Docker boot loop: container restarts after exit code 0

3 answers

Add your answer or comment

mcp-memory-service Docker boot loop: container restarts after exit code 0

3 answers

Add your answer or comment

Related questions