MCP Server constantly restarting

I’m running this inside a digital ocean installation in Docker using Roo Code - have been using the perplexity MCP fine.

Loaded it up fine, and once I put the env variables in, the chat worked well with AI responses. I was able to start the MCP Agent Service - but when I’m in Roo on my MCP Server screen, it loads for about .5 second and then shows…Processing request of type ListToolsRequest Processing request of type ListResourcesRequest Processing request of type ListResourceTemplatesRequest

I was able to screen capture what it shows for half a second and it clearly shows it’s connecting but the AI in my coding environment thinks that the MCP server is restarting so effectively getting thrown straight out I think…then it shows whats on the left of the image

Anyhow would be much appreciated. I’m very keen to use this and think that Cole has done a great job - just want to use it!

Here is some updated info from my AI Assistant:

Technical Details

  1. The Archon services are running properly in Docker containers:
  • Main Archon container (archon) is running and stable
  • The archon-mcp container keeps restarting or creates new instances
  1. Each time I click “Retry Connection” in the MCP settings panel, it creates a new Docker container instead of reusing the existing one, leading to multiple instances of archon-mcp running simultaneously.
  2. The MCP settings configuration I’m using is:
{
  "mcpServers": {
    "archon": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "--network=localai_default",
        "-e",
        "GRAPH_SERVICE_URL",
        "archon-mcp:latest"
      ],
      "env": {
        "GRAPH_SERVICE_URL": "http://host.docker.internal:8100"
      },
      "disabled": false,
      "alwaysAllow": []
    }
  }
}
  1. When checking Docker containers with docker ps | grep archon, I can see multiple archon-mcp containers being created, with the original one in a “Restarting” state."

1 Like

Could you first try completely wiping all the current Archon MCP containers and trying again? I haven’t seen this happen before and I’m not sure what error in the MCP container could cause it to constantly restart!

If that doesn’t help I could see if I could reproduce the issue

@ai52-coder

I have seen this behavior as well and it seemed to be related to running multiple docker instances. I ran into an instance where I had closed the docker desktop but the daemon did not shutdown. I was not able to see the duplicate process in Docker Desktop but in the cli using the Docker ps. Once I cleared both lists (as @ColeMedin suggests above) the issued was resolved.

1 Like