Clean Systemd files on exit #792
Conversation
|
Thanks for your PR,
To skip the vendors CIs, Maintainers can use one of:
|
| signal.Notify(sigc, syscall.SIGTERM) | ||
|
|
||
| errChan := make(chan error) | ||
| defer close(errChan) |
There was a problem hiding this comment.
I used the method used here to close channels. But I believe there is a race in general, on exit (once main goes away, or if os.Exit is called) the goroutines are killed off immediately. So the goroutines listening on this channel will never get the chance to run their intended actions.
|
cc @e0ne |
Pull Request Test Coverage Report for Build 11341895365Details
💛 - Coveralls |
souleb
force-pushed
the
fix-systemd-cleanup
branch
from
9700ce2
to
1f68fa8
Compare
If implemented all systemD related files will be cleaned up on exit.
souleb
force-pushed
the
fix-systemd-cleanup
branch
from
1f68fa8
to
f13f0ab
Compare
| @@ -207,6 +207,14 @@ func (dn *Daemon) Run(stopCh <-chan struct{}, exitCh <-chan error) error { | |||
| for { | |||
| select { | |||
| case <-stopCh: | |||
| // clean files from host if we are running in systemd mode | |||
| if vars.UsingSystemdMode { | |||
| err := systemd.CleanSriovFilesFromHost(vars.ClusterType == consts.ClusterTypeOpenshift) | |||
There was a problem hiding this comment.
IIUC when the daemon is stopped we delete systemd files.
if the node reboots (given pods get stop signal) wont it delete the files ? which in turn mean the node will not have SR-IOV configured on startup ?
There was a problem hiding this comment.
What you say is correct. We should have a way to identify whether a reboot has been requested. In that case the cleanUp has to be a best effort because we can have an ongoing reboot and also a spec change in the daemonSet.
Signed-off-by: Soule BA <[email protected]>
We have run into an issue where removing the operator leaves behind the systemd files.
If implemented all systemD related files will be cleaned up on exit on non
openshiftdistributions.