[Nut-upsuser] Tripp-Lite OMNI750ISO

[David Vree]

On 03/25/2013 09:36 AM, Charles Lepple wrote:
> On Mar 25, 2013, at 9:20 AM, David Vree wrote:
>
>> [ups.delay.shutdown]
>> Interval to wait after shutdown with delay command (seconds)
>> Type: STRING
>> Value: 64
>>
>> There is nothing apparently settable about its LB point via the driver.  BTW -- What do you think "ups.delay.shutdown" means?
> ups.delay.shutdown is the number of seconds between when the driver (in response to LB) sends the UPS the shutdown command, and when the UPS should actually pull power.
You mean pull power to the server?  Then I should definitely increase 
that from 64 to 180 seconds as it takes some minutes to shut to 20+ VMs.
> The alternative, if you are not as concerned about maximizing runtime, is to start a timer with upssched when the UPS first loses power, and cancel the timer if the power returns before the LB threshold is hit. I would advocate the belt-and-suspenders approach of doing both, just in case your battery starts to weaken, and you hit the LB threshold before the timer expires. The upssched approach has the disadvantage of being time-based rather than percentage-based, but it should get the job done.
I will look into this approach as well.  Thanks for the pointer.