9.2.1.6 power fail-shutdown

[tomelgato]

Installed a ups connection to a apc smart ups 1500 via ap9617 management card.

Everything works fine, except a nice surprice during power failure shutdown:

--- snip ---
Broadcast Message from root@freenas.local
(no tty) at 17:37 CEST...

UPS ups on battery


Broadcast Message from root@freenas.local
(no tty) at 17:41 CEST...

Executing automatic power-fail shutdown
--- snap ---

… but doesn't shutdown

Config:
freenas 9.2.1.6 on hp NL54 with 16gb ram
APC Smart Ups 1500 w. AP9617 Network management card (latest firmware)

Driver used (UPS Mode: Master): various ups3 (snmp-ups)


Any ideas?

 

[dlavigne]

Did you figure this out? If not, does typing shutdown from the FreeNAS shell properly shutdown the system or result in the same behavior?

 

[tomelgato]

No, didnt investigate further without feedback. Im more on the linux then bsd side ;)


Shutdown from the shell does work normal, also from the GUI.


Where are the shutdown scripts?

 

[enemy85]

Have u waited 4 minutes more after the second message?

 

[tomelgato]

Not sure but i think so....

Why a 4 minute delay? This messes up the shutdown strategy, especially when there are more than 1 machine is running on the same ups.

 

[enemy85]

U have to consider how works the shutdown service...
Let's suppose u put 300 in the box (5minutes), then u plug the power cord off the wall and let the system run on battery. After 5 minutes u will get the "executing power fail shutdown" message, BUT only after OTHER 5minutes the system will actually start the shutdown procedure...
Try and see if it works like this

 

[tomelgato]

yes .. but...... ;)

you loose 5 min of runtime in this case and u create a blind window of 5 minutes: if you have a short power failure (which happens here more often) then u start the shutdown 5 min more early ... if power returns during the "wait for whatever shutdown needs 5 minutes" timeframe i assume the box still shuts down.

will check out the extra delay on we.

 

[enemy85]

i'm not saying that is correct how it works...i'm just saying how it's designed to! (someone designed it to work this way...)
Actually i don't either know what happens if the power comes back during the "second 5 minutes"... if you try let me know!