Describe the bug
After running the Doris MCP Server for an extended period, it inevitably becomes unusable and throws errors for all requests. I attempted to recover it using docker restart doris-mcp-server, but this did not restore functionality. The only way to get it working again is to run docker-compose down followed by docker-compose up -d.
This indicates that docker restart is not sufficient to clear whatever internal state is causing the failure, likely related to stale database connections or internal server resources.
To Reproduce
Run the Doris MCP Server using Docker Compose for an extended period (several hours/days) under normal load.
Wait until the server starts throwing errors (e.g., connection timeouts, internal server errors).
Run docker restart doris-mcp-server and test if the server is functional again. (It will likely still be broken).
Run docker-compose down followed by docker-compose up -d.
The server will be functional again.
Expected behavior
docker restart doris-mcp-server should be sufficient to clear any internal state and restore the service to a healthy state, similar to a full container recreation.
Environment (please complete the following information):
doris-mcp-server version: [请填写你的版本，例如 0.6.0]
Deployment Method: Docker / Docker Compose
OS: [例如 Ubuntu 22.04]
Additional context
This is very similar to issues reported with database connection pools timing out in long-running sessions . It seems like docker restart might not be effectively resetting the internal connection pool or async event loop, leaving it in a broken state. A complete container teardown (down) is required to truly reset the environment. This could be related to the async event loop issues fixed in earlier versions that still cause problems with internal resource cleanup .
Describe the bug
After running the Doris MCP Server for an extended period, it inevitably becomes unusable and throws errors for all requests. I attempted to recover it using docker restart doris-mcp-server, but this did not restore functionality. The only way to get it working again is to run docker-compose down followed by docker-compose up -d.
This indicates that docker restart is not sufficient to clear whatever internal state is causing the failure, likely related to stale database connections or internal server resources.
To Reproduce
Run the Doris MCP Server using Docker Compose for an extended period (several hours/days) under normal load.
Wait until the server starts throwing errors (e.g., connection timeouts, internal server errors).
Run docker restart doris-mcp-server and test if the server is functional again. (It will likely still be broken).
Run docker-compose down followed by docker-compose up -d.
The server will be functional again.
Expected behavior
docker restart doris-mcp-server should be sufficient to clear any internal state and restore the service to a healthy state, similar to a full container recreation.
Environment (please complete the following information):
doris-mcp-server version: [请填写你的版本，例如 0.6.0]
Deployment Method: Docker / Docker Compose
OS: [例如 Ubuntu 22.04]
Additional context
This is very similar to issues reported with database connection pools timing out in long-running sessions . It seems like docker restart might not be effectively resetting the internal connection pool or async event loop, leaving it in a broken state. A complete container teardown (down) is required to truly reset the environment. This could be related to the async event loop issues fixed in earlier versions that still cause problems with internal resource cleanup .