Open colleenXu opened 1 month ago
There's a limit to the logs that can be retrieved due to the way the TRAPI logging system is managed. We could attempt to 'salvage' logs in the event of a TimeoutError
but there wouldn't be a very solid way to ensure they were in order, and we'd still be missing quite a bit.
So....this is maybe something we can't do? Should this issue be closed?
When the execution of a template times out, I notice only a few TRAPI logs from this execution and the TimeoutError.
It would be helpful to have all the normal TRAPI logs up to the Timeout for debugging. Ex: so we have a sense of what happened - did execution get stuck somewhere / sub-queries take a while / NodeNorm take a while? How far did the template execution go?
Example of what this looks like: