RE: FU: UPS on an i5 520 - Not Your Father's Attachment

>       We are running 2 520s here with the 1827/serial cable 
> setup and we have tested the items you mention after point 
> #3.  We have successfully received the messages that occur 
> when the UPS in in bypass mode and when the UPS goes on/off 
> battery (utility power lost/utility power restored).
> We have tested for those messages and received them and also 
> have received them in a 'live' situation when we had utility 
> power problems.  In addition to that... If I unplug the cable 
> while the system is up and running, it will tell me so... 
> we've tested it (in fact, I just walked in to our development 
> box and yanked the cable/reattached it and got CPI0961 and 
> CPI0962.  If you're not getting messages for those types of 
> events, it sounds like something is still wrong with your 
> setup... it sounds like maybe the signal output to your UPS 
> might still be suspect.

You're right.  I stand corrected.  It's the signal from the UPS that
changed.  That same cable (old UPS) gave the attached, on bypass, on
battery, and low battery.  But when switching to the new UPS, the same port
and cable now only give the attached and battery low signal.  There's no way
I can use the new UPS with the i5 serial connection and get the protection I
want.  I have to use the J14 port with this UPS.

I must be getting too old to handle this many variables anymore. <g>
 
> With that said, I agree that IBM implementation needs some 
> work (it was a MESS when the 520 hit the streets)... I think 
> it's stemming from the fairly new existence of the FSP and 
> firmware PTFs and all those concerns.  I had something odd 
> occur last week when we loaded some firmware PTFs and then 
> when they finally applied at IPL I noticed a message about 
> the UPS no longer being attached.  It required another power 
> down, unplug for 5-10 mins, and IPL to sense the cable was 
> reattached again.  The support person didn't know why this 
> occured from loading MH* PTFs, but it did.  They are hurting 
> their system reliability and wasting some of those minutes of 
> availability throughout the year that add up quick when 
> people start talking about five 9's and such. ;)

I agree.  In handling this whole situation there were so many times I didn't
know whether I was afoot or horseback.  I was initially under the impression
that the UPS end of things was identical to before (same port, same cable,
and the UPS people assuring me it was the same).  The fact that the i5
connection was what was known to be different, coupled with the CE telling
me IBM _knew_ there were firmware issues that the MH* PTFs were to fix made
me absolutely convinced the _whole_ problem was on the i5 side.  Ultimately
in the end, I found there were problems on both sides.  I got the new UPS
when I still had my 270.  I'll bet I had some of the same issues with the
270 attached to the new UPS and never knew it.

I now monitor the UPS in 2 ways:

1) Monitoring the message queue and handling those messages programmatically
(which I always did).

2) A job scheduler job wakes up every 15 minutes, uses the MATMATR API, and
reports the results.  If there are no issues, a QHST message says so.  If
there _are_ issues, I get notified all over the place.  The possible
'issues' are a) no UPS attached, b) on bypass, c) on battery, d) low
battery, and e) UPS message queue is not being monitored.

That's probably overkill, but I'm gunshy right now.