Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Major performance drop between 0.8.7 and 0.9.0 #1073

Open
MRigal opened this issue Jul 17, 2015 · 3 comments
Open

Major performance drop between 0.8.7 and 0.9.0 #1073

MRigal opened this issue Jul 17, 2015 · 3 comments

Comments

@MRigal
Copy link
Member

MRigal commented Jul 17, 2015

I've just tried to pass my production servers from 0.8.7 (and some subclasses) to 0.10.0 and had to rollback. I've tried with 0.9.0 as a provisory solution but it was exactly the same.

Some changes for the 0.9.0 introduced some serious performance drop, only noticeable at a certain load. We had a huge increase in running processes and CPU usage coming from the system and not from the user, in parallel to load increase (increasing exponentially with the load, behaving normal at low level).

Just creating this ticket to remember to take a deeper look at performance. I've been through the diff between 0.8.7 and 0.9.0 and tried to make a branch rolling back some of the changes, without success for now...

If somebody is interested I think this effort could benefit to many people.

@touilleMan
Copy link
Member

@thedrow Personally, I choose Python3 greatness over Pypy speed... but I also gave my 50 bucks to support th, Pypy3 port project, can't wait for it ! ^^

@MRigal MRigal self-assigned this Aug 18, 2015
@MRigal
Copy link
Member Author

MRigal commented Aug 18, 2015

OK, I've spent quite a bit on time on it and I could reduce the amount of possibilities to this diff. v0.8.7...be3643c
It happened quite at the beginning of the work on the 0.9.0 branch. Unfortunately, it is not easy to isolate the problem, since it is the time where Ross made a big dev effort over two days, but thus merged branches in all directions.

They aren't so many critical parts, so I'll try to cherry-pick the best parts and isolate the problem. It's neither so easy to test since it shows up only with significant load and only after a while, as if there were some problems on the clean-up part.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants
@wojcikstefan @MRigal @touilleMan and others