Hi Daniel,
   We had this very same problem last week.  It seemed to come out of nowhere.  
We are still in 'discovery' mode with z/VM and Linux on our z900, so I didn't 
think it was a cpu problem (low usage), but I suspected a paging problem so I 
added another paging volume,...then another,..but the problem persisted.   The 
page space utilization percentage just kept growing. 

The other Listserv responses to your problem are right on.  IBM suggested we 
change the SRM parms, particularly the STORBUF parm. Ours SRM parms were set to 
the default. 
We bumped the parm up to "STORBUF: Q1=500% Q2=400% Q3=300%", and our lock-out 
problem went away. Paging is cruising at 23%, instead of 80%.    We are looking 
at the other tuning parms in SRM, as IBM suggested. 

Gabe Torres - Systems Programmer
State of Nevada 
Dept of Information Technology - Computing Div. 
 This communication, including any attachments, may contain confidential 
information and is intended only for the individual or entity to whom it is 
addressed. Any review, dissemination or copying of this communication by anyone 
other than the intended recipient is strictly prohibited. If you are not the 
intended recipient, please contact the sender by reply e-Mail and delete all 
copies 



-----Original Message-----
From: Linux on 390 Port [mailto:[EMAIL PROTECTED] Behalf Of
Daniel Jarboe
Sent: Monday, November 22, 2004 4:53 AM
To: [EMAIL PROTECTED]
Subject: VM dispatching


A few weeks ago we had what over time grew into a big problem
dispatching guests.  Guests would stop being dispatched sometimes for 10
minutes at a time, for no reason that we could identify.  Usually
shutting some guests down would get everyone dispatching again, but it
was a mess.  We applied some VM PTF's (like UM30888) and things got much
better, but there are still times early in the morning where our
performance guy's accounting records show the guests getting zero cycles
in a minute interval or two.

For example, last Wednesday a few guests got zero cycles (averaged over
minute intervals) at 00:23:00-00:25:59, and this morning it was 4 guests
from 00:21:00-00:23:59 (specifically linintra 00:21:00-00:23:59, linsmtp
00:21:00-00:22:59, linwsm 00:21:00-00:22:59, linps1 00:22:00-00:23:59).

An IND Q shows that we now have non quick dispatch guests running in
queues other than Q3, which is different than from before the PTFs (we
are on z/VM 4.3 Guestlan).

IND Q
TCPIPII       Q0 PS  00000419/00000084 AI00032       Q1 R00
00001165/00001144
TCPIP         Q0 R   00001054/00000367 LININTRA      Q2 PS
00012923/00012885
LINPROXY      Q0 PS  00015499/00015461 LINSMTP       Q1 PS
00012242/00012183
LINFS1        Q3 PS  00045610/00053128 LINPS2        Q1 PS
00004611/00004544
LINPOP        Q3 PS  00006052/00005455 LINFS3        Q3 PS
00024126/00038166
LINFS2        Q3 PS  00054785/00065536 LINUX1        Q3 PS
00006535/00007603

Occasionally we will see these messages in linux syslog:

Nov 17 00:25:14 linpop kernel: addrconf_verify(): too slow; jiffies -
now = 63

We run the guest's storage pretty tight on the linux side; our largest
three images are 256MB (the samba file servers), with VDISK allocated as
needed (our most active one right now has another 256MB split over 4
VDISKs).  Our smallest guest is 32MB.

Are there any VM commands we can use to get a better idea why VM might
stop dispatching some of the guests temporarily, or any other
ideas/suggestions anyone can offer?

Thanks,
~ Daniel

----------------------------------------------------------------------
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

----------------------------------------------------------------------
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

Reply via email to