Details
-
Bug
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
-
None
Description
There are known kernel bugs with the CFS scheduler resulting in processes being stuck and unable to be killed after an OOM, leading to cgroups::destroy() looping. Touching one of the control files, e.g., `echo 19000 > cpu.cfs_quota_us` can unstick it.
Attachments
Issue Links
- is duplicated by
-
MESOS-2421 Processes can be stuck in D state and prevent container destroy
- Resolved