pokey@well.UUCP (Jef Poskanzer) (05/20/89)
Since we switched from a Vax to a Balance 8, the ps command has been
unacceptably slow. As in many minutes for a simple ps -x. Everything
else runs nice and fast, so what's the deal? Has anyone else come
across this problem?
---
Jef
Jef Poskanzer jef@helios.ee.lbl.gov ...well!pokey
"I am a man of high calibre; that is to say, a large bore."pwolfe@kailand.KAI.COM (05/22/89)
> /* Written by pokey@well.UUCP in kailand:comp.sys.sequent */ > Since we switched from a Vax to a Balance 8, the ps command has been > unacceptably slow. As in many minutes for a simple ps -x. Everything > else runs nice and fast, so what's the deal? Has anyone else come > across this problem? No. The output from the command "ps -x" starts displaying within a second on our Balance. It uses a file in /tmp (.ps_data), is it possible that your /tmp is located on a very slow and/or very busy disk? Look at the output of "pstat -T", and check the number of max processes. Maybe your process table is too large (changed in kernel config parameters). Check out the latest SURF proceedings for Ron Parsons article on tuning Dynix V3.0 systems. Besides being entertaining, it contains oodles of useful information about measuring and tuning system performace. Patrick Wolfe (pat@kai.com, {uunet,uiucuxc,sequent}!kailand!pat)
hanst@maestro.htsa.aha.nl (Hans Trompert) (05/22/89)
In article <11749@well.UUCP> Jef Poskanzer <jef@helios.ee.lbl.gov> writes: >Since we switched from a Vax to a Balance 8, the ps command has been >unacceptably slow. As in many minutes for a simple ps -x. Everything >else runs nice and fast, so what's the deal? Has anyone else come >across this problem? We have a Balance 8 with 16mB RAM and 8 CPU's, our load average is normally around 0.5. A ps doesn't take more than 1 sec. Here's the ouput of a 'time ps -x': PID TT STAT TIME COMMAND 5500 ic Z 0:00 <defunct> 220 ic I 0:09 -msh (msh) 838 ic I 0:25 sysline -pjbDhim 839 ic S 0:14 -msh (msh) 4516 ic I 0:02 rn 4536 ic I 0:00 /bin/sh -c Pnews -h /user/staff/hanst/.rnhead 4537 ic I 0:02 Pnews /usr/local/bin/Pnews -h /user/staff/hanst/.rnhead 4570 ic S 0:04 /usr/ucb/vi /user/staff/hanst/.article 5501 ic S 0:00 sh -c time ps -x 5502 ic S 0:00 awk { 5503 ic R 0:00 pstat -T 5504 ic S 0:00 time ps -x 5505 ic R 0:00 ps -x 0.8 real 0.1 user 0.5 sys I don't know what is wrong with your ps, but it isn't normal. -- )!&@%^$!@&(^*$*!&@^%$()!@&#$*)&(^A%()*&()*^%*()&!@$(*&@(!%*)&(*&@%*(&%(*&!%&*(* & Hans Trompert # Disclaimer: % | Algemene Hogeschool Amsterdam ~ I don't know, you can shoot ^ \ Technische en Maritieme Faculteit ) me in the partylights ! # % E-mail: hanst@maestro.htsa.aha.nl ! (partylights not included) ) !@&*&^%_)@!%*@!&)^*&@$&^%!_(@$+~_@$+_)@!($+)_@!%$&@$(~)!@&+@%+)@!&+%)$(@+*&()$+
andy@syma.sussex.ac.uk (Andy Clews) (05/23/89)
From article <2400053@kailand>, by pwolfe@kailand.KAI.COM: > Check out the latest SURF proceedings for Ron Parsons article on tuning Dynix > V3.0 systems. Besides being entertaining, it contains oodles of useful > information about measuring and tuning system performace. Can someone tell me from where I can obtain this article from the UK? Has it been posted to the net, or indeed, can it be? Thanks -- Andy Clews, Computing Service, Univ. of Sussex, Brighton BN1 9QN, ENGLAND JANET: andy@syma.sussex.ac.uk BITNET: andy%syma.sussex.ac.uk@uk.ac Voice: +44 273 606755 ext.2129
pwolfe@kailand.KAI.COM (05/26/89)
/* Written by andy@syma.sussex.ac.uk in kailand:comp.sys.sequent */ > From article <2400053@kailand>, by pwolfe@kailand.KAI.COM: > > Check out the latest SURF proceedings for Ron Parsons article on tuning Dynix > > V3.0 systems. Besides being entertaining, it contains oodles of useful > > information about measuring and tuning system performace. > > Can someone tell me from where I can obtain this article from the UK? > Has it been posted to the net, or indeed, can it be? That particular article is 50 printed pages long, about half text and half graphics. Too large to type, and too costly to make copies and mail them to all interested. I suggest you try to get a copy of the whole proceedings. There are other interesting items in there too. SURF, for those who don't know, stands for the Sequent User's Resource Forum, the name of the offcial Sequent user's group. If you don't already belong, you should. You aren't required to attend the yearly conference, but if you can, it's a wonderful way of getting questions answered, learning from other people's experiences, and finding out about new software and hardware products. Even if you don't go, you get a copy of the conference proceedings. I think that you join SURF (and hopefully get a copy of last year's proceedings) by calling your local Sequent sales representative or the Sequent main offices at (800-854-0428 or 503-626-5700), and ask for the SURF representative. If someone else knows a better way, please post. Patrick Wolfe (pat@kai.com, {uunet,uiucuxc}!kailand!pat) System Manager, Kuck and Associates, Inc. "Inix! Unix! We all scream for Dynix!"