MCP FAQ

Error code

Description

Troubleshooting

11200044 - MCP_CONNECTION_REFUSED

The client's connection to the target service was refused. Possible causes:

• The service is not running.

• The port is not open or is blocked by a firewall.

• The network is unreachable or the address is configured incorrectly.

• Run curl <service address> to test the connectivity.

• Check the downstream service logs to confirm that the MCP service is running.

• Check the gateway or firewall configuration of the downstream service for connection-rejection rules, such as an IP address whitelist.

11200045 - MCP_CONNECTION_TIMEOUT

The connection timed out. Possible causes:

• High network latency or packet loss.

• The server is under high load and responding slowly.

• Unstable network route or connection.

• Long connection path when accessing through a proxy.

Retry the connection 2-3 times:

• If the connection is occasionally successful, the issue is likely network-related.

• If the issue persists, check the gateway or firewall configuration along the downstream service path for rules such as an IP address whitelist.

11200046 - MCP_REQUEST_TIMEOUT

The connection was established, but the server did not respond in time. Possible causes:

• The server is taking too long to process the request or lacks sufficient resources.

• Slow response time due to network bandwidth limitations or high latency.

Retry the request 2-3 times. If the issue persists:

• Break down the business logic to reduce the processing time of a single request, or implement asynchronous processing.

• For npx/uvx deployments: Enable logging in the Function Compute (FC) Console to view detailed error information. You can also switch to Express Mode to reduce cold start times.

11200047 - MCP_NETWORK_ERROR

An error related to network reachability, DNS resolution, or a dropped connection.

Possible causes:

• The domain name cannot be resolved or is resolving incorrectly.

• The connection was closed by the remote host or an intermediate device.

• The target host is unreachable or there is a routing issue.

Retry the request 2-3 times. If the issue persists:

• Run nslookup to verify that domain name resolution is working correctly. If an error occurs, change the DNS and retry.

• Check the url field in the configuration, and run curl <url> to test if the service responds.

11200048 - MCP_SSL_ERROR

The TLS/SSL handshake or certificate validation failed. Possible causes:

• The server certificate has expired, the domain name does not match the certificate, or the certificate chain is incomplete.

• The client does not trust the server certificate (for example, a self-signed certificate).

• The TLS version or cipher suite is incompatible.

• A proxy or gateway has modified the certificate.

• Open the service address in a browser to check if the certificate has expired or if the domain name does not match.

• If you are using a self-signed certificate, ensure it is configured to be trusted.

• Disable any proxies and test with a direct connection to determine if a proxy is modifying the certificate.

11200049 - MCP_SERVER_HTTP_UNAUTHORIZED

HTTP 401. The request lacks valid authentication credentials or the credentials have expired. Possible causes:

• Authentication information (such as a token or API key) was not provided or was not passed correctly.

• The token has expired, been revoked, or is malformed.

• The server requires specific request headers or cookies that were not included.

• Consult the documentation for the downstream MCP service to understand its authentication method.

• Add the correct authentication information to the MCP service, such as the Authorization header.

• Use a tool like curl to test a direct connection to the downstream service and ensure the configuration is correct.

11200050 - MCP_SERVER_HTTP_FORBIDDEN

HTTP 403. The request was authenticated, but the user does not have permission to access the resource. Possible causes:

• The current account or role does not have permission to access the MCP or instance.

• Access is restricted by IP address or source.

• Consult the downstream MCP service documentation to understand its authorization requirements.

• Use a tool like curl to test a direct connection to the downstream service and ensure your credentials and permissions are correct.

11200051 - MCP_HTTP_RATE_LIMIT

HTTP 429. The request rate has exceeded the defined limit. Possible causes:

• Too many requests were sent in a short period.

• The quota for the current account or instance has been exhausted.

• Reduce the call frequency. If the response includes a Retry-After header, wait for the specified interval before retrying.

• If your application requires a higher quota, contact the MCP service provider to request an increase.

11200052 - MCP_HTTP_CLIENT_ERROR

An HTTP 4xx client error other than 400, 401, 403, 404, 405, or 429. Possible causes:

• The request method, URL, headers, or body do not meet the server's requirements.

• The request body is too large or the URL is too long.

• The server timed out while waiting for the request (for example, HTTP 408).

• Examine the specific status code and error message in the response body and consult the downstream MCP service documentation.

• Use the curl tool to test and view the complete request and response, as well as the downstream MCP service logs.

11200053 - MCP_HTTP_SERVER_ERROR

An HTTP 5xx error, indicating an internal server or gateway error. Possible causes:

• The MCP service or gateway is experiencing an issue, is overloaded, or is under maintenance.

• A dependent downstream service is unavailable.

Retry the request 2-3 times. If the issue persists:

• Check the response error details and the downstream MCP service logs.

• For npx/uvx deployments: View the runtime logs in the Function Compute (FC) Console (you must enable logging first).

• Use the curl tool to test, and view the complete request and response and the downstream MCP service logs.

11200054 - MCP_PROTOCOL_ERROR

The response could not be parsed according to the MCP protocol. Possible causes:

• The server returned a standard JSON-RPC error (for example, Parse error, Invalid Request, Method not found, Invalid params, or Internal error).

• The protocol or version is incompatible.

• Correct the request format, method name, or parameters based on the JSON-RPC error code.

• Confirm that the type in the configuration matches the endpoint path: "sse" corresponds to /sse, and "streamableHttp" corresponds to /mcp.

11200055 - MCP_SESSION_NOT_FOUND

The session does not exist or has expired. Possible causes:

• The client is still using an old session after the server has restarted or the session has expired.

• The session ID was passed incorrectly or is missing.

• A server routing policy sent the request to a different instance.

• Optimize the downstream MCP service to comply with standard MCP protocol behavior.

• Check the logs. The downstream MCP service may have restarted unexpectedly, causing the session to be lost.

11200056 - MCP_UNKNOWN_ERROR

An error that does not fit any of the other categories. The original error message is included in the response. Possible causes:

• An unclassified error originating from the runtime environment or its dependencies.

• Investigate based on the original error message and the context.

• For npx/uvx deployments: Enable logging in the Function Compute (FC) Console to view runtime logs.

11200057 - MCP_INIT_TIMEOUT

The server did not send a ready signal during the initialization phase. Possible causes:

• The server did not push initialization data in time.

• High network latency or packet loss.

• An intermediate proxy or gateway does not properly support long-lived connections.

Retry the connection 2-3 times. If the issue persists:

• Go to MCP Management to check the access address and the transmission method (sse or streamableHttp).

• Run curl <endpoint> to test the connectivity.

• If using a reverse proxy, enable long-lived connections and set a sufficient timeout period.

• For npx/uvx deployments: Check the startup logs in the Function Compute (FC) Console. Basic Mode has a cold start delay; switch to Express Mode for faster startup.

11200058 - MCP_SERVER_HTTP_METHOD_NOT_ALLOWED

HTTP 405. The endpoint does not accept the HTTP method used in the request. Possible causes:

• The transport method does not match the access path (for example, using SSE with a message endpoint).

• Using an unsupported method for a URL (for example, using GET instead of POST).

• Confirm that the type in the configuration matches the endpoint path: "sse" corresponds to GET /sse, and "streamableHttp" corresponds to POST /mcp. If the configuration is incorrect, modify it in MCP Management and then redeploy.

• We recommend using a standard MCP SDK to implement the downstream MCP service.

11200059 - MCP_SERVER_HTTP_NOT_FOUND

HTTP 404. The requested path does not exist. Possible causes:

• The access URL path is configured incorrectly (for example, a missing path prefix).

• The server's router or gateway is not forwarding requests for this path to the MCP service.

• The instance is offline or was not deployed correctly.

• Make sure that the type in the configuration matches the endpoint path: "sse" for GET /sse, and "streamableHttp" for POST /mcp. If the configuration is incorrect, modify it in MCP Management and then redeploy.

• Go to MCP Management to check the service status. If the service is not deployed, redeploy it.

• Use the curl tool to test the connectivity.

11200060 - MCP_SERVER_HTTP_BAD_REQUEST

HTTP 400. The request format or parameters are invalid. Possible causes:

• The request body is not valid JSON or does not comply with the MCP/JSON-RPC format.

• Required parameters are missing or have the wrong type.

• Request headers or query parameters failed server-side validation.

• Check if the field types and required fields in your request match the server's requirements.

• Use the curl tool to test connectivity.