[pve-devel] [PATCH ha-manager 4/4] tell systemd to restart CRM and LRM on failures

[Thomas Lamprecht]

On 09/13/2016 07:44 AM, Dietmar Maurer wrote:
> I wonder if you run into situations where this is useful? if so, can
> you describe an example scenario?
>

This was more thought as a exception against an not exact specifiable 
scenario.
I saw it in the systemd docs and as it, AFAIS, cannot introduce 
regressions but
may help if the service crashes (havok running OOM, or a unknown bug 
triggering bad stuff) I added it.

>> +Restart=on-failure
> Or do you just think/feel this is useful?

Rather this, yes :) on-abnormal would be maybe an better option, this 
would no restart it on "unclean exit codes".
A few other HA related service files, e.g. the one from corosync(-qnetd) 
have it.