Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
None
-
None
-
None
Description
Seen this while testing locally, when the slave was started with "./bin/mesos-slave.sh". Since the slave has a controlling TTY attached to it, as soon as it dies all the underlying cgroup executors get a signal and die. This doesn't happen with process isolator because we setsid() after fork(). We need to do the same in cgroups isolator.