Another side effect btw is that some of our log files (we add a file handler to the logger) end up at 0 bytes. This specifically happens with Ray and ClearML and does not reproduce locally
We just inherit from logging.Handler and use that in our logging.config.dictConfig ; weird thing is that it still logs most of the tasks, just not the last one?
Some commits related to subprocesses and thread handling 🙂
What do you mean 😄 Using logging.config.dictConfig(...)
Hi UnevenDolphin73 , which clearml version are you using?
I thought so too - so I added flush calls just in case, but nothing's changed.
This is somewhat weird since it always happens in the above scenario (Ray + ClearML), and always in the last task/job from Ray
Well, the thing is ClearML also uses dictConfig, and I think you might be overriding its settings...
I believe it is maybe a race condition that's tangent to clearml now...
Example configuration -version: 1 disable_existing_loggers: true formatters: simple: format: '%(asctime)s %(levelname)-9s %(name)-24s: %(message)s' filters: brackets: (): ccutils.logger.BracketFilter handlers: console: class: ccmlp.utils.TqdmStreamHandler level: INFO formatter: simple filters: [brackets] loggers: # Set logging levels for specific packages urllib3: level: WARNING matplotlib: level: WARNING botocore: level: WARNING fsspec: level: WARNING s3fs: level: WARNING boto3: level: WARNING s3transfer: level: WARNING git: level: WARNING ray: level: WARNING PIL: level: WARNING root: level: DEBUG handlers: [console]
I'll try with 1.1.5 first, then 1.1.6rc0
Might very well be - do you touch other handlers?
SuccessfulKoala55 could this be related to the monkey patching for logging platform? We have our own logging handlers that we use in this case
Or do you mean the contents of the configuration, probably :face_palm: ... one moment
If that's the case, wouldn't it apply across the board? This happens in a single task within ray - the other tasks (I have many in a single run) are fine