Since b46f13ef113e690493364dabc87821699626d8e6 in zaza, if running tests
on an OpenStack cloud provider, each bundle test concludes with a
cleanup routine that needs the python-openstackclient package in
requirements.txt for the tox environment, plus the OS_ environment vars
exported/passed through such that the Openstack client actually works.
If these are not set, the test fails on the first bundle cleanup and
does not continue to the remaining bundles.
On systems where net namespaces have an ID associated, the output of
ip netns list includes that information. This change updates the means
to collect the list of namespaces to just include the names.
When a kubernetes pod has a readonly mount defined,
the mountpoint is alerted by nrpe check_ro_filesystem
mount, but is mounted as described by the deployment.
Ignore /run/containerd backend paths in
check_ro_filesystems by default
nrpe_utils.py: ensure permission for /var/lib/nagios
This nrpe charm will install nagios-nrpe-server deb package.
In its preinst script[0], it will add nagios user and create `/var/lib/nagios` as home dir.
When other charm like hw-health relates to this charm, they will:
1) setup cronjob to run script
2) generate output file into /var/lib/nagios
3) read output file from /var/lib/nagios
Before, these are all done via root user, so no permission issue.
But recently, the cronjob user is switched from root to nagios, which caused following issues:
1) original output file was created by root, cronjob script by nagios user can not write to it.
2) In some situation, owner of `/var/lib/nagios/` is changed to root, cronjob script can not write file into this dir.
1) ensure `/var/lib/nagios` is owned by nagios user
2) setgid on dir group, to ensure any new created file in `/var/lib/nagios` has group `nagios`.
NOTE: this patch avoids to chown recursively, which implies, if there is a legacy output file
owned by root, we have to fix it manually, or fix it from the related charm side.