Hi all -
Just an FYI that I have enabled a watchdog timer (virtual) device in our
virt-install command for new VMs in staging, a few days ago. This is
something Kevin asked me to look into a while back.
There shouldn't be any noticeable changes, but as staging VMs get
rebuilt, they will have a /dev/watchdog device and a 'watchdog' service
running. Relevant commit is here: .
Let me know if there are any problems or questions or reboot loops
(hopefully not!) on new VM builds. Existing (non-rebuilt) VMs do not
have the watchdog device at this time, so this message only concerns new
or rebuilt VMs.
Also, if anyone is really familiar with the watchdog daemon and can
suggest other checks to add, please chime in! It looks like it is
capable of a fair number of types of checks, but right now I just have
it in the most simple configuration possible.
If this goes well with stg VM rebuilds for a while, we can drop the
conditionals and move it to prod in a similar fashion (as VMs get
rebuilt). Alternatively, we could edit our VMs XML files and add the
device manually, and install watchdog (via ansible) on everything. I
defer to Kevin for the decision on that.