It seems that `/var/log/rook` inside the VM does not contain any files.
Getting the logs from the Pods through kubectl may not be as stable, but
it should get some logs when minikube is still/again available.
Signed-off-by: Niels de Vos <ndevos@redhat.com>
More details of the Rook (and Ceph) deployment should be useful when
troubleshooting CI failures. This now includes the status of the most
important Kubernetes objects, and all the logs Ceph stores on the host.
Updates: #1969
Signed-off-by: Niels de Vos <ndevos@redhat.com>
Without the `-w` argument, the output of `top` gets truncated, and the
commandline of the processes is not comlete. It would be useful to eb
able to tell which command uses 100% CPU in an output like:
17377 root 20 0 110.8m 8.2m 0.0 0.1 0:00.89 S `- containerd+
17414 167 20 0 1036.7m 59.6m 0.0 0.4 0:03.47 S `- ceph-o+
40875 root 20 0 283.9m 30.4m 100.0 0.2 0:00.23 R `- ceph
Updates: #1969
Signed-off-by: Niels de Vos <ndevos@redhat.com>
The new `system-status.sh` script logs the status of the host and the
minikube VM. This gets executed when a CI job fails, and should aid in
troubleshooting spurious failures.
Updates: #1969
Signed-off-by: Niels de Vos <ndevos@redhat.com>
