You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I cannot tell CPython that memory is limited to value X. I run thousands of python workers @mesos stack where I need to pre slice the memory. That is big operation issue with python.
It will be killer feature to have something similar to "xmx" parameter in java world. I know that pyston won't free needed memory but it should be aware of max memory and make more pressure on GC work.
The text was updated successfully, but these errors were encountered:
That's a really interesting idea! What would happen if you approach the memory limit? I would imagine that there is usually relatively little memory for the cycle collector to free since most memory management is done via refcounting, but maybe there are other space-time tradeoffs that could be adjusted.
What would happen if you approach the memory limit?
in java world GC is fired much more often + doing more careful work
In general python world has a lot of memory quirks (such [1]) - for my OPS point of view: runtime that is aware of memory limits would be large step forward. Right now we loose http requests while CPython is killed by mem limit on cgroup and I have no way to tell if it was: memory leak or legit memory allocation since "maybe GC was about to fire but didn't manage to" (because of oom kill).
I cannot tell CPython that memory is limited to value X. I run thousands of python workers @mesos stack where I need to pre slice the memory. That is big operation issue with python.
It will be killer feature to have something similar to "xmx" parameter in java world. I know that pyston won't free needed memory but it should be aware of max memory and make more pressure on GC work.
The text was updated successfully, but these errors were encountered: