Re: Instance won't start up after powerfail

Ed Stevens <nospam_at_noway.nohow> wrote in message news:<t2nh90pqcvju3nr7dkddup2o5p5k8h8119_at_4ax.com>...
> On 4 May 2004 16:16:00 -0700, joel-garry_at_home.com (Joel Garry) wrote:
>
> >Stan Brown <stanb_at_panix.com> wrote in message news:<c78kfk$o67$1_at_reader2.panix.com>...
> >> In <4096cc22$0$32558$afc38c87_at_news.optusnet.com.au> "Howard J. Rogers" <hjr_at_dizwell.com> writes:
> >>
> >> >Stan Brown wrote:
>
> >> >> In <c75lsu$p5g$05$1_at_news.t-online.com> "Dorian Büttner" <dorian.buettner_at_gmx.de> writes:
> >> >>
> >> >>
> >> >>>is it data file 6
> >> >>>or '/db_local3/oradata/pwhse/tables/trend.dbf'
> >> >>>?
> >> >>
> >> >>
> >> >>
> >> >> For the record, that was it.
> >> >>
> >> >> Thanks to everyne!
>
> >> >Also for the record, Richard's original advice was 100% correct for
> >> >version 8.0 and upwards (file number alone is sufficient there).
> >> >Clearly, version 7 (which, like me, he probably doesn't have a museum
> >> >capable of running) was a bit different!
>
> >> >Glad you got it sorted this time Stan without hassle! (PS... ever
> >> >thought of buying a UPS??)
> >>
> >> It's on a 75KVA industrial UPS, which has 2 feeds from 2 different buses,
> >> and it's powered by the 125VDC station bateries that control the switchgear
> >> in the same location.
> >>
> >> But hardware cannot overcome incompetence in people. Putting the UPS in
> >> "bypass", and ifnoring the "UPS not in sync" alarm on the computer screen,
> >> and opening random breakers without authiriztion can overcome even the best
> >> hardware.

Sounds like one gov't shop I was in, big secruity/blast doors and the whole bit. No one ever 'fessed up to doing it, but they sure tightened up sign-in routines after that! :-O

> >>
> >> Srry, I'm stilll upset about this!
> >
> >http://groups.google.com/groups?selm=6dnn80%24q7h%241%40pebble.ml.org&oe=UTF-8&output=gplain
> >
> >And of course, the loudest sound I ever heard, including '70's rock
> >concerts, was a 1 Farad capacitor:
> >http://groups.google.com/groups?selm=68wNsB1w165w%40netlink.cts.com&oe=UTF-8&output=gplain
> >
> >I'm _still_ laughing about it.
> >
> >jg
>
> Did I read that right? 1 Farad? It's been a loooong time since I
> studied that stuff ... (like, in jr. high in the early '60s when I was
> interested in ham radio and you pretty much had to roll your own) ...
> but aren't capacitors usually measured in micro- and pico-farads?
> That would make a 1 Farad .... crikeys, I don't even want to think
> about it!

Yes, IIRC there were four of them, each the size of a classic Quaker Oats box. The whole thing was about a 5 ft cube. The person who ordered it was expecting something breadbox sized, and was surprised when half-a-dozen marines showed up pushing the thing in. Hecho en Mexico, where it had to go back to be fixed.

>
> But reading the story made me think about our whole approach to UPS
> protection in the data center -- for which I am decidedly *NOT*
> responsible for. In this case, the server in question would have been
> much better served sitting on a little APC unit costing a couple of
> hundred bucks. By extension, I wonder if it wouldn't be less
> expensive up front as well as more flexible and more powerful in the
> long run to eliminate the single point of failure, massive data center
> wide UPS systems and just put each server on its own little unit.

When you starting adding up a pile of servers, it makes sense to get something bigger and fancier. For example, I've got 3-soon-to-be-4 PC's surrounding me now for OAS ("my" pc and three tiers), and the Back_UPS sitting there basically does one of the dual power supplies of two of the boxes, my pc, the terminal, the port-switch, and a linksys. Barely enough power to shutdown cleanly. Since the server room UPS has about 45 minutes for all the stuff it has now, obviously need more UPS power to move all these things in there. You do get an economy of scale with UPS, not to mention higher quality with the more pro units. Especially a couple years down the road when all those little units start losing batteries at random times.

I am also glad I am not responsible for such things.

>
> I'm also reminded of the data center that had a big diesel powered
> backup system that they tested religiously by running the data center
> on it one weekend every month. Then, when they actually lost utility
> power and went to start the backup they realized that the electric
> starter motor ran on .... you guesed it!

So the biker-dude sysadmin hooked his Harley up to it and... :-)

jg

--
@home.com is bogus.
http://www.banknotes.com/AQ1.JPG