Re: [U2] Universe spooler daemon memory and cpu usage

Wed, 15 Jun 2005 22:48:28 -0700

Also check the size of the usplog reflection in the UVSPOOL directory (>CONFIG ALL to get the installed pathname). In addition to Dan's suggestion, you could also be chewing up a lot of CPU scanning a huge memory based spooler queue (usplog is a disk copy of the in-memory queue). This is not unusual if you generate many, many hold files. On some platforms, the size of the in-memory version of usplog will grow to the point to where UniVerse crashes. There is no UniVerse solution to this other than smack people about the head and shoulders and insist they clear out their old hold files (not to be confused with &HOLD& files).
<AD> You can contact me off-list at [EMAIL PROTECTED] for information about my spooler product that solves both of the above problems. </AD> 


--

Regards,

Clif

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
W. Clifton Oliver, CCP
CLIFTON OLIVER & ASSOCIATES
Tel: +1 619 460 5678    Web: www.oliver.com
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


On Jun 15, 2005, at 9:37 PM, Dan Fitzgerald wrote:


transposing from aix & other unices, what's lpstat say? Look for a 
problem communicating with a printer.





"Our greatest duty in this life is to help others. And please, if you 
can't help them, could you at least not hurt them?" - H.H. the Dalai 
Lama
"When buying & selling are controlled by legislation, the first thing 
to be bought & sold are the legislators" - P.J. O'Rourke

Dan Fitzgerald





From: "Dan Barrett" <[EMAIL PROTECTED]>
Reply-To: [email protected]
To: <[EMAIL PROTECTED]>
Subject: [U2] Universe spooler daemon memory and cpu usage
Date: Thu, 16 Jun 2005 11:31:31 +0930

Hi everyone,


I recently installed UniVerse 10.0.20 onto a Solaris 8 box 
(Enterprise 450,
2 x 400Mhz Processors, 1GB memory.) along with Redback, to run a 
hybrid

Pick/ASPX/Uniobjects based banking application.


Everything seems to work Ok, however the universe spooler daemon 
process

"usd" seems to chew up an excessive amount of memory and CPU time.

I've reset the spooler daemon but there seems to be no effect, I 
haven't
tried stopping and restarting universe or the spooler daemon yet, as 
this is

now a production banking system.

eg - here's output from "top"  -

load averages:  1.99,  2.07,  1.95               holden
11:21:55
596 processes: 592 sleeping, 1 zombie, 2 stopped, 1 on cpu

CPU states: 64.1% idle, 16.9% user, 16.7% kernel,  2.3% iowait,  0.0% 
swap

Memory: 1024M real, 224M free, 936M swap in use, 1.2G swap free


   PID USERNAME THR PR NCE  SIZE   RES STATE   TIME FLTS    CPU 
COMMAND

   391 root       1  0   0 55.0M 38.9M sleep  82.5H    0 20.87% usd
  5314 ultracs    1 23   0 12.2M 6304K sleep 152:09    0  2.23% uvsh
  5320 ultracs    1  4   0 12.3M 6408K sleep 151:40    5  2.22% uvsh
 11880 ultracs    1 13   0 11.9M 7264K sleep   1:10    3  2.21% uvsh
  5319 ultracs    1 22   0 12.2M 6288K sleep 149:23    0  2.10% uvsh
  8451 lrestas    1 11   0 11.0M 6264K sleep  10:35    0  1.34% uvsh
  2532 ultracs    1 24   0 13.0M 7152K sleep  87:12    3  1.13% uvsh
 25447 morgan     1 34   0 2720K 1080K sleep  92:44    0  1.11% top
  2516 ultracs    1 33   0 13.1M 7200K sleep  82:31    0  0.92% uvsh
  2536 ultracs    1 34   0 13.8M 7880K sleep  85:04    0  0.86% uvsh
  2529 ultracs    1 35   0 12.9M 7024K sleep  84:53    0  0.86% uvsh
  2530 ultracs    1 33   0 13.8M 7976K sleep  86:35    0  0.85% uvsh
  2541 ultracs    1 36   0 13.1M 7328K sleep  89:14    0  0.84% uvsh
  2512 ultracs    1 46   0 13.0M 7128K sleep  86:07    0  0.80% uvsh
  2519 ultracs    1 34   0 13.1M 7240K sleep  80:17    0  0.77% uvsh

  5319 ultracs    1 33   0 12.3M 6344K sleep 149:20    0  0.77% uvsh


You can see "usd" tops the list at 20% CPU time, and 55MB memory. 
Memory

usage has been gradually creeping up, about 1 MB a day or so.

Has anyone come across this before? I'm not a universe expert by any 
means

having come from a mostly oracle background, but this seems pretty
excessive.

Printing still works as expected but users tell me the system does 
take a
reasonable time to generate a hold file, ie 3 seconds or so for a 
small
file. On earler version of universe this was basically instantaneous 
on the

same platform.


The app we're running does generate a reasonable number of print jobs 
and
spooler hold files, but the level of spooler daemon cpu and memory 
usage

doesn't drop, even when we're not printing anything.

Any ideas?
Thanks

Dan
-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/

-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/

-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/






















					Reply via email to