Do you have Glance, or SOS, or any other performance measurement tool ?
If you do, did you check disk queue lengths before/after ?
If you did, did they change ? How much ? Up or down ?
At least, that could help narrow the problem.
HTH,
Christian "have passport, will travel"
> -----Message d'origine-----
> De : HP-3000 Systems Discussion [mailto:[log in to unmask]] De la part
> de Gordon Montgomery
> Envoyé : lundi 26 février 2007 21:19
> À : [log in to unmask]
> Objet : [HP3000-L] CPU seconds in job question
>
> Y'all have been so helpful with my recent adventures, that
> I've got another puzzle for you.
>
> When I rebuilt my system this past week, I took the opportunity
> to move from 6.0 to 7.0. I had the new tape sitting aroung for
> many years since, "if it's not broken, don't fix it..". Things
> seem to be running just fine, however I have a couple of
> weekend jobs whose execution times increased, some
> significantly. In looking at the print outs, the CPU seconds
> have greatly increased. In one job the CPU seconds increased
> from 6299 last week to 26446 this week. What could account
> for this kind of increase? I have not recompiled any of the programs
> yet. I think that will be my next step, but is it really necessary?
>
> The user volume used to sit on two SCSI cards, each with
> 15 4gig drives, and used Mirror/iX. Now the user volume sits
> on one SCSI card connected to an HP Autoraid with 6 LUNS
> defined at 20gig each. Would a disk bottle neck account for
> that much more CPU overhead?
>
> In case you haven't guessed, I know just enough to be dangerous,
> so thanks for any insight you could give.
>
> Gordon Montgomery
> Living Scriptures, Inc.
> (801) 627-2000
> [log in to unmask]
>
> * To join/leave the list, search archives, change list settings, *
> * etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
|
