server: remove reaper, let runc take care of reaping #272
Conversation
Signed-off-by: Antonio Murdaca <[email protected]>
|
I didn't see any error either, Good work!! I'll revert that commit from my PR. |
|
The reaping isn't handled by runC -- it's not possible to have a subreaper setup if you use |
right, either way, this patch does seem to fix the issue - @cyphar wdyt? |
| "github.com/Sirupsen/logrus" | ||
| "github.com/kubernetes-incubator/cri-o/utils" | ||
| "github.com/containernetworking/cni/pkg/ns" | ||
| "github.com/kubernetes-incubator/cri-o/utils" |
There was a problem hiding this comment.
Just a nit: This change isn't really needed.
There was a problem hiding this comment.
this seems a gofmt problem, I have create a commit to add a check in CI.
There was a problem hiding this comment.
Mmm pretty sure I'm running gofmt with vim, will double check, thanks
There was a problem hiding this comment.
@runcom yeah, I mean you are right in this PR, this is a gofmt problem brought before.
|
LGTM |
2 similar comments
|
LGTM |
|
LGTM |
This is likely fixing tests and
wait: no child processeserror when runningcri-o.Fix #268
@mrunalp @jjlakis @feiskyer @Crazykev please test this out (as I cannot reproduce
ECHILDerror anymore with this patch)runc does already take care of reaping childs (as I read the code) - I believe we're late here in to reap and killing/stopping/removing/state causes ECHILD so frequently when ops are really fast.
Signed-off-by: Antonio Murdaca [email protected]