AbortError: This operation was aborted MCP SSE

[isaacraja]

This is happening often with the bigquery mcp server. I couldn't find a way to reproduce it reliably - But happens when more than two people connect to librechat. I'm using api gateway and ec2 instance in aws.
Error processing tool: "query_mcp_bigquery" tool call failed: This operation was aborted.
The tool returned the data and then librechat sends an
abort. I used supergateway to to convert existing stdio server to SSE server and I am also using this particular https://github.com/ergut/mcp-bigquery-server/ that is written in JavaScript.

I read couple of discussions and issues and i saw a mention of timeout in api gateway - Not sure what value should i change.

docker compose logs mcp-bigquery-sg | [supergateway] Child → SSE: { mcp-bigquery-sg | result: { content: [ [Object] ], isError: false }, mcp-bigquery-sg | jsonrpc: '2.0', mcp-bigquery-sg | id: 28 mcp-bigquery-sg | } LibreChat-API | {"level":"debug","message":"[MCP][User: 12345] Updated last activity timestamp: 2025-05-13T06:24:47.132Z","timestamp":"2025-05-13T06:24:47.132Z"} LibreChat-API | {"level":"debug","message":"[MCP][User: 12345][bigquery] Transport sending: {\"jsonrpc\":\"2.0\",\"method\":\"notifications/cancelled\",\"params\":{\"requestId\":21,\"reason\":\"AbortError: This operation was aborted\"}}","timestamp":"2025-05-13T06:25:38.351Z"}

I've cache set to false in librechat.yaml(if it matters)

[danny-avila]

I used supergateway to to convert existing stdio server to SSE server

This might be the root of your issue. The conversion might enhance the connectivity to your MCP server but it might still have the drawback of sharing the stdio process. This would need to be tested outside of LibreChat.