Is there a tool/service that can automatically and perpetually instrument and profile python server processes? I'm thinking about processes like Celery or RQ workers? I'd like to get method-level performance timers averaged across multiple similar job executions.
New Relic will do this for Celery but it only has experimental support for RQ. Unfortunately it's not recommended for short lived tasks, and we have lots of those.
I'm aware of cProfile and line_profiler, but I'm hoping to find a service I can use in production where I don't have to capture the output and aggregate it myself. While a permanent service/tool would be preferred, if there's a tool that will aggregate the output of multiple cProfile runs that might work as well.
BTW the processes are running on Heroku non-web worker dynos.
Aucun commentaire:
Enregistrer un commentaire