Why am I receiving 500 errors when connecting to MCP servers in Pipedream's developer mode and could this occur in production?

Help
1

This topic was automatically generated from Slack. You can find the original thread here.

On Pipedream developer mode, sometimes, when connecting to mcp servers, we get 500 errors like this:

breChat-API | 2025-06-24 19:21:06 error: [MCP][User: 6831a77a46d7304e714d8248][pipedream-linkedin] Connection failed: Error POSTing to endpoint (HTTP 500): {“jsonrpc”:“2.0”,“error”:{“code”:-32603,“message”:“Internal server error”},“id”:null}
LibreChat-API | 2025-06-24 19:21:09 error: [MCP][User: 6831a77a46d7304e714d8248][pipedream-linkedin] Connection failed: Error POSTing to endpoint (HTTP 500): {“jsonrpc”:“2.0”,“error”:{“code”:-32603,“message”:“Internal server error”},“id”:null}
LibreChat-API | 2025-06-24 19:21:13 error: [MCP][User: 6831a77a46d7304e714d8248][pipedream-linkedin] Connection failed: Error POSTing to endpoint (HTTP 500): {“jsonrpc”:“2.0”,“error”:{“code”:-32603,“message”:“Internal server error”},“id”:null}
LibreChat-API | 2025-06-24 19:21:13 error: [MCP][User: 6831a77a46d7304e714d8248][pipedream-linkedin] Failed to connect after 3 attempts Error POSTing to endpoint (HTTP 500): {“jsonrpc”:“2.0”,“error”:{“code”:-32603,“message”:“Internal server error”},“id”:null}
LibreChat-API | 2025-06-24 19:21:13 error: [MCP][User: 6831a77a46d7304e714d8248][pipedream-linkedin] Failed to establish connection Error POSTing to endpoint (HTTP 500): {“jsonrpc”:“2.0”,“error”:{“code”:-32603,“message”:“Internal server error”},“id”:null}
LibreChat-API | 2025-06-24 19:21:13 error: [MCPInitializer][UserMCPController.connectMCPServer] Failed to connect single MCP server ‘pipedream-linkedin’ for user 6831a77a46d7304e714d8248: Error POSTing to endpoint (HTTP 500): {“jsonrpc”:“2.0”,“error”:{“code”:-32603,“message”:“Internal server error”},“id”:null}

It looks like the error is not on my side. Is this because of the developer mode that the servers are unstable or can this also happen on production?

2

It shouldn’t have anything to do with development vs production, I’m not sure exactly what those errors are. What MCP client are you using? How are you making those calls?

3

Issue resolved - it was on our side!

We’re using the @modelcontextprotocol/sdk TypeScript client with StreamableHTTPClientTransport to connect to your servers.

The HTTP 500 errors weren’t actually coming from your endpoints. We had a bug in our connection logic that was:

  1. Artificially throttling ping requests - throwing “Empty result” errors if pings happened within 5 minutes
  2. Using a singleton pattern that caused connection conflicts when multiple users tried to initialize their MCP servers simultaneously

The “HTTP 500” errors were actually our transport layer reporting these self-inflicted connection failures, not real server errors from
your side.

Thanks for the quick response - the servers are working perfectly.