[GitHub] [mesos] cf-natali edited a comment on pull request #388: Fixed a bug where the cgroup task killer leaves the cgroup frozen.

[GitBox <gi...@apache.org>]

cf-natali edited a comment on pull request #388:
URL: https://github.com/apache/mesos/pull/388#issuecomment-863608013


   @qianzhangxa Did you manage to reproduce it?
   
   The updated MR only handles the discard at the beginning of a freeze/kill/thaw cycle, therefore ensuring we don't leave a cgroup frozen.
   Obviously it means that discard will be ignored while the kill/thaw is in progress, but I think it's what we want since interrupting them would be quite bad, and in practice kill is non-blocking and thaw shouldn't block forever.
   
   I test this by:
   - running all the tests
   - running the failing test in a loop, adding logging to check that the discard code path was exercised (and `freezeTimedout`)
   - I ran a test continuously starting new tasks, using `strace --inject` as explained above to change timings and make timeouts more likely.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org