[tech] [committee] Auction TOMORROW for UPS

[Harry McNally]

On 09/03/11 17:57, Daniel Axtens wrote:
>>>> There's an auction coming up TOMORROW at Ross's[1], where some new 3kVA 
>>>> Liebert PS3000RT-230 UPS's are going to be auctioned. If possible, I would 
>>>> like to get one of these for the club, as well as an extra battery unit. 
>>>> They are 2RU rack mount with 15 plugs - which will mean that we would be 
>>>> buying it with no guarantee that we can power it. Mark Tearle tells me the 
>>>> addition of another 15A socket in the machine room will cost around $500.
> 
> I know we're massive fans of uprecords here at UCC, but I don't know if spending $700-$900ish to prevent power dropouts is a value for money for the club.
> 
> Having said that, what is the likely impact of power fluctuations on the lifespan of our equipment? 

Hi Daniel

That is a point to consider but we have a dinky little UPS on each CPU box
(only, not screens) here and the main benefit is carrying the gear across very
short disturbances. The UPS jumps in, runs on battery for it's obligatory 5
seconds until the power looks stable again, and returns the supply to mains.

What is saves in hassle more than compensates for the little bit of power it
uses. I think if the Ross' bid can secure the UPSes and extended battery boxes
(buyers for them alone are less likely) then, even if only the main unit was
re-stocked with batteries (and the other left unused) the 6 minute run time
will cover the main situation of very short disturbances.

We've set up nut here (tho I haven't done that yet for the replacement
CompuGuard 650 I purchased the other day) so the system runs 90 seconds on
battery and shuts down. Nut can also communicate across multiple machines
running on one (or several) UPS to sequentially shut down systems. I selected
90 seconds because it was more about preserving file system integrity (and
file recovery from vim etc) than trying to run the system across the outage.
This gets back to your original lifespan question; it's not about uprecords
for us.

For the UPS' sake, running only for 90 seconds doesn't take the battery
anywhere near to the UPS' "discharged battery" threshold which, for these
small SLA batteries, kills them without too many cycles to flat. I suspect the
UPS manufactures underrate the battery and are relying on the idea that most
power losses are short.

My 2c
Harry