[
https://issues.apache.org/jira/browse/MESOS-8515?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16346199#comment-16346199
]
Qian Zhang commented on MESOS-8515:
-----------------------------------
As a fix, we can set the executor pid for the container in
`DockerContainerizerProcess::_recover`, but it will cause the test
`DockerContainerizerTest.ROOT_DOCKER_SlaveRecoveryTaskContainer` failed. It
might be because we kill the executor process with `os::killtree` in
`DockerContainerizerProcess::destroy` (without the fix, we will not do it since
the container's executor pid is `None()` after recovery).
> Docker containerizer does not recover the executor pid
> ------------------------------------------------------
>
> Key: MESOS-8515
> URL: https://issues.apache.org/jira/browse/MESOS-8515
> Project: Mesos
> Issue Type: Bug
> Components: docker
> Reporter: Qian Zhang
> Priority: Major
>
> During agent recovery, Docker containerizer does not recover the executor
> pid, so each `Container.executorPid` will be `None()` after the recovery.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
