scripts/ci: create test cgroup controllers once#1780
scripts/ci: create test cgroup controllers once#1780avagin merged 2 commits intocheckpoint-restore:criu-devfrom
Conversation
The idea that each zdtm.py should have own helder, so that two zdtm.py that are running on the same host don't effect each other. Fixes: checkpoint-restore#1774 Signed-off-by: Andrei Vagin <[email protected]>
zdtm.py mounts two named controllers for tests. In CI, we run zdtm.py a few times, so we can mount (create) these controllers once to avoid any unwanted effects. Signed-off-by: Andrei Vagin <[email protected]>
adrianreber
left a comment
adrianreber
left a comment
There was a problem hiding this comment.
I always had a suspicion that it might be something like this, but I was able to reproduce it locally without any other zdtm process running in parallel.
From what I have seen it is less about two zdtm running in parallel and more about a cgroup being lazily removed from the kernel and while zdtm re-creates the cgroup the kernel still removes the previous cgroup.
This sounds like something that could help and so far CI is happy.
LGTM as it seems to provide a fix for the problem we are seeing.
Yes, it looks like this and it is why the second patch is here. The case when we create and remove cgroup controllers back and forth isn't common in real life. Usually, controllers are created once and are never destroyed. |
2 participants
zdtm.py mounts two named controllers for tests. In CI, we run zdtm.py a few
times, so we can mount (create) these controllers once from the ci script to
avoid any unwanted effects.
Fixes: #1774