-
Bug
-
Resolution: Incomplete
-
Critical
-
None
Jenkins version: 1.538
In jenkins slave node, there are a lot of ssh-agent process linger in the system which has older timestamp than the one for slave java process.
it means terminating jenkins slave process does not kill those ssh-agent process.
ssh-agent process:
jenkinsuser 31771 1 0 Jul17 ? 00:00:00 ssh-agent -s
the unix socket file taken by this process:
ssh-agent 31771 jenkinsuser 3u unix 0xffff8810596ef380 0t0 16634661 /tmp/ssh-VzHtq31768/agent.31768
The file does not actually exist.
ls -l /tmp/ssh-VzHtq31768/agent.31768
ls: cannot access /tmp/ssh-VzHtq31768/agent.31768: No such file or directory
If jenkins is not responsible to cleanup those lingering ssh-agent process after slave java process is re-started, should jenkins admin to do the cleanup?