Unattended, unintended system shutdown - why?

[David Fletcher]

On Sat, 2015-03-07 at 23:58 +0000, Dan Purgert wrote:
> On Sat, 07 Mar 2015 22:49:09 +0000, David Fletcher wrote:
> 
> > On Sat, 2015-03-07 at 22:22 +0000, Dan Purgert wrote:
> >> On Sat, 07 Mar 2015 13:14:35 +0000, David Fletcher wrote:
> >> Could be that the battery's not as reliable as you thought -- how old
> >> is it?
> > 
> > Batteries replaced 2014-03-07
> > 
> 
> Do the batteries self-test OK?  I've had one or two banks that fresh from 
> APC (well, after recommended charge cycle) didn't pass the tests.  Turned 
> out they had bad cells. 

Don't know what the hell was going on there but it wasn't good - I also
run my desktop system off the same UPS so to save standby power I have
the PC, monitor, speakers, and scanner connected to UPS power via an IEC
switched power block. I ssh'd into the server from the desktop, stopped
apcupsd, the sudo apctest. Selected option 2 I think it was to perform a
self test and I can't imagine how or why but the switch in the power
block started spitting blue sparks. That's never good. I killed
everything with the button on the UPS and switched off the power block,
then pushed the button on the UPS to start it going again.

Switched the power block on again and booted up the desktop. Tried to
ssh into the server and nothing - no route to host - ping - nothing.
Bugger. Pulled the power plug from the server and tried again. Nothing.
More expletives. Plugged in the VGA cable from the monitor and swapped
source, and the server was stuck at the grub screen. Even more
expletives. Plugged in a USB keyboard to which the server responded and
allowed me to select the Ubuntu option. It then booted Ubuntu and fired
up as normal. ping ans ssh working. I did a sudo shutdown -h now and it
shut down. Pulled and reinserted the power plug again and this time it
booted right up without intervention, as it's supposed to.

Tried switching off the power to everything at the wall socket, the UPS
fired up and started beeping while the server and desktop kept right on
running. Switched the wall socket back on, the beeping stopped, and the
server logged the power outage....

Last few log file entries follow:-
2015-03-08 10:58:08 +0000  apcupsd exiting, signal 15
2015-03-08 10:58:08 +0000  apcupsd shutdown succeeded
2015-03-08 10:59:22 +0000  apcupsd 3.14.10 (13 September 2011) debian
startup succeeded
2015-03-08 11:14:12 +0000  Power failure.
2015-03-08 11:14:18 +0000  Running on UPS batteries.
2015-03-08 11:14:19 +0000  Mains returned. No longer on UPS batteries.
2015-03-08 11:14:19 +0000  Power is back. UPS running on mains.

So, I don't know exactly what that self test does in addition to
checking the batteries but my try out indicated to me that it's pretty
nasty and does not encourage me to try it again....

Dave


> 
> 
> > 
> > 
> >> In addition to that - have you changed your system layout since getting
> >> the battery?  E.g. newer / more power hungry server, or other devices
> >> getting plugged into the battery?
> > 
> > Only things changed are the hard drives. Complete mystery to me.
> 
> Definitely.  I didn't see anything in the log entries myself, but I admit 
> that I'm hardly an expert at reading them.
> 
>