In some specific cases, high intensive operations sent through SQL can take a long time to process resulting in a request timeout message similar to the one below:
[Kinetica][Server] (81321) Error during GetResponses: Error sinking: Endpoint:bulk_update_internal, Error: Request timed out after 1226.55 seconds (config param timeout set to 1200 seconds), Chunk: 32of 33, Endpoint:bulk_update_internal (TM/Cc:95); code:1 'Error' (TM/Cc:65); code:1 'Error' (S/Sc:57); code:1 'Error' (S/Sc:57); code:1 'Error'.
While this error appears to indicate the request has failed, this error actually means that the SQL client used has timed out waiting for a response from the server, but id doesn't mean that the server stopped processing the request.
Affects Versions: All
The request timeout can be increased when running long executing queries either at the client side or at the server side:
- Server side:
- Navigate to /opt/gpudb/connectors/odbcserver/bin
- Edit file gpudbodbc.ini $ vi gpudbodbc.ini
- At the bottom of the file change the Timeout property to the desired amount (in milliseconds)
** Any change to the ODBC server configuration requires a restart of the odbc server
- Client side:
- In the driver properties set the Idle Timeout to a higher number (In seconds)
- In the connection string add ConnectionTimeout=xx (Where xx is in seconds)