It might be interesting to log the activity for jobs from the controller node, this will help us catch errors that happen before the dendro processes run inside a successfully started job.
For example, I just got a Batch job failing right before the start, some authorization issue when pulling the docker image... the Dendro logs are empty, but I can find the source of error using boto3:
It might be interesting to log the activity for jobs from the controller node, this will help us catch errors that happen before the dendro processes run inside a successfully started job.
For example, I just got a Batch job failing right before the start, some authorization issue when pulling the docker image... the Dendro logs are empty, but I can find the source of error using boto3: