Open BhautikDonga opened 3 weeks ago
Please find the related graphs and extra information from this discussion: https://discuss.ray.io/t/memory-leakage-with-ray-put/15501
if Is it because your producers are producing faster than your consumers are consuming
if Is it because your producers are producing faster than your consumers are consuming
I have tested with check of queue size ( updated reproduction script above ) and the memory leakage is still there.
Hi @BhautikDonga,
While we investigate why there is memory leak, I'd strongly recommend you to use Ray Data library (given you are doing data processing) which handles things like back-pressure for you and can avoid these memory leaks.
@jjyao Thanks for looking into this issue. I will go through the Ray Data library and see how I can move to that. It would be great if you can suggest or share some docs for how easily I can move to Ray Data library.
One more thing we have analysed is that, this memory leakage is somehow related to https://github.com/ray-project/ray/issues/47274.
Please let me know, if I can provide any other informations.
What happened + What you expected to happen
I am seeing the memory leakage while putting objects (numpy array) into object store via
ray.put()
.Versions / Dependencies
I am using
ray==2.34.0
andpython3.10
withubuntu20.04
.Reproduction script
Issue Severity
High: It blocks me from completing my task.