NO NO NO NO, PLEASE NO
Anybody who raises LDUBUF really needs to look at their paging devices AND know
if users
are loading. Raising LDUBUF is one of the WORST things you can do for
performance - it
either does nothing, or allows you to page to death. The ONLY documented
performance
analysis of LDUBUF in last 20 years shows that lowering LDUBUF stopped a system
from
thrashing. And if you think a system that does not page can possibly be
impacted by
LDUBUF, you need, really really need to go to the Velocity Software performance workshop.
(Sorry Detro, maybe your company doesn't allow you to take classes?)
If your overcommit ratios is above 1, then it's your STORBUF - this problem in Linux is
ALWAYS STORBUF. So you are actually doing the reverse of what you should be doing if you
raise LDUBUF and not STORBUF. Add up your virtual machine sizes. What is that
as compared to what your real storage is? Just set STORBUF to 300 300 300.
REALLY.
There is a performance class in september, found at
"VELOCITYSOFTWARE.COM/seminar/workshop.html" I would recomend you at least see
day one.
Horst, you should look at the handout from the configuration guidelines you received from
me about 3 years ago. Or review "VELOCITYSOFTWARE.COM/PRESENT/CONFIG".
Stahr, Lea wrote:
Thanks Gary. I am using the default LDUBUF : Q1=100% Q2=75% Q3=60% and will try
this, then the STORBUF: Q1=140% Q2=130% Q3=120%.
Lea Stahr
zVM, Linux and zLinux Administrator
Navistar, Inc.
630-753-5445
[EMAIL PROTECTED]
________________________________
From: Linux on 390 Port [mailto:[EMAIL PROTECTED] On Behalf Of Gary Detro
Sent: Thursday, August 07, 2008 8:29 AM
To: LINUX-390@VM.MARIST.EDU
Subject: Re: AW: Linux sles9 felt in coma
here is what I use for a linux system
LDUBUF : Q1=200% Q2=200% Q3=200%
STORBUF: Q1=300% Q2=200% Q3=200%
It is probably your LDUBUF for Q3 users that is causing it. Change that one 1st and see how it does.
Thanks,
Detro
Senior IT Specialist
1177 S. Beltline Rd, Coppell, TX 75019
Internal Mail Stop: 71-01-3001O
Phone: 469-549-8174 (t/l603)
Fax: 469-549-8149 (t/l 603)
[EMAIL PROTECTED]
Global Solution Center <http://w3.ibm.com/support/stss/gesc.html> (GSC)
<http://www.ibm.com/ondemand>
From:
"Rempel, Horst" <[EMAIL PROTECTED]>
To:
LINUX-390@VM.MARIST.EDU
Date:
08/07/08 08:26 AM
Subject:
AW: Linux sles9 felt in coma
________________________________
Hello Jonathan,
I can see in perftk that LX004 was in CL3,ELIG while sleeping.
Here are my srm-settings:
q srm
IABIAS : INTENSITY=90%; DURATION=2
LDUBUF : Q1=100% Q2=75% Q3=60%
STORBUF: Q1=125% Q2=105% Q3=95%
DSPBUF : Q1=32767 Q2=32767 Q3=32767
DISPATCHING MINOR TIMESLICE = 5 MS
MAXWSS : LIMIT=9999%
...... : PAGES=999999
XSTORE : 0%
Ready; T=0.01/0.01 15:15:02
I think these are the default settings.
Are there any recommendations for srm settings for this VM. We are mainly
running linuxguests with databases.
I do not think that this is a storage problem, because we normaly use only
10-20% of the available swapspace and have no VM paging.
There are no quickdsp setting in this VM.
kind regards
Horst Rempel
-----Ursprüngliche Nachricht-----
Von: Linux on 390 Port [mailto:LINUX-390@VM.MARIST.EDU
<mailto:LINUX-390@VM.MARIST.EDU> ] Im Auftrag von Quay, Jonathan (IHG)
Gesendet: Donnerstag, 7. August 2008 13:07
An: LINUX-390@VM.MARIST.EDU
Betreff: Re: Linux sles9 felt in coma
Look back in the performance toolkit data and see if an eligible list formed
around the time LX004 hung up. You may need to add more paging exposures
and/or adjust your SRM.
________________________________
From: Linux on 390 Port on behalf of Rempel, Horst
Sent: Thu 8/7/2008 6:49 AM
To: LINUX-390@VM.MARIST.EDU
Subject: Linux sles9 felt in coma
Hello,
one of our Linux sles9 SP4 DB2 databaseservers running under z/VM 5.3.0
RSU0801 felt yesterday afternoon in a kind of coma.
The linux stopped running. The putty-session got an timeout. I was no longer
able to ping.
The processor for this lpar was only 20% busy for other linux guests at this
time.
After that I tried to logon directly to the green screen. But I only got a
black screen with LOGON LX004 at the top.
The ind user lx004 shows no I/Os, no cpu consumtion. The performance toolkit
displayed only zeros for this LX004.
After force/xautolog LX004 everything worked fine again.
Looking in the linux logs gave no hint what lead to the coma situation.
There is enough space free in the filesystems.
I have no idea what happened yesterday.
Has anybody an idea ? What can I do if the situation come back again ?
This mail is crossposted to LINUX-390@VM.MARIST.EDU and [EMAIL PROTECTED]
kind regards
Horst Rempel
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
begin:vcard
fn:Barton Robinson
n:Robinson;Barton
adr;dom:;;PO 390640;Mountain View;CA;94039-0640
email;internet:[EMAIL PROTECTED]
title:Sr. Architect
tel;work:650-964-8867
note:If you can't measure it, I'm just not interested
x-mozilla-html:FALSE
url:http://velocitysoftware.com
version:2.1
end:vcard