failed systemctl services?

[stan]

On Mon, 04 May 2015 16:40:49 +1000
"Michael D. Setzer II" <mikes at kuentos.guam.net> wrote:

> I was checking on systemctl status and found some that were listed as 
> failed?
> One was caused since the vboxdrv modules for 3.19.5 kernel were not
> being installed by yum for some reason. They didn't show up in the
> list, but was able to locate and manually install it to correct that
> error.

Sounds like a packaging error.  You should probably open a bugzilla
against either the kernel or vboxdrv.

> The others I'm not sure of. Network-wait-online would report failed
> after boot on some systems, but running a systemctl restart would fix
> the issue, ended up changing the timeout from 30 to 90, and reboots
> seem to result in no failures. 
> 
> On a few machines got udisks2.service failed with Cannot allocate
> memory message, but a restart on that also resulted in success. 

This sounds like timing issues.  IIRC, systemd uses threaded processing
in order to speed up the boot process.  Lots of opportunity for failure
there:  race, blocking, timing, etc.

> Is there a process that automatically will try to restart failed
> systemctl services, or would it be worth creating something to do so?
> Perhaps in cron.hourly? 

I'm no expert, but my understanding is that there is not.

> Might be these are harmless, but not sure.

It doesn't seem harmless that services that the user expects to be
started, and are configured properly, fail.

It would be good if you opened a bugzilla against systemd for an RFE
requesting such a utility.