> -----Original Message-----
> Behalf Of Rob van der Heij
> Sent: Friday, February 13, 2009 1:54 PM
> 
> It's very hard to fill in the blanks. As long as you drive 
> the cmm_pages from VM (which is imho the only place where you 
> have relevant metrics to provide the proper setting) you will 
> have to remember what you did. 

If you want to drive cmm_pages according to current memory/storage
restraint then yes, VM is the only place with relavant metrics. VM does
on the onther hand not know much about what happens inside the guest
(cmm2 not considered). Why shouldn't the guest voluntarily give up most
free memory and file cache above e.g. 300 MB (of course depending on the
workload)?

>Though Leland referred to an 
> APAR, I think it was a change in the kernel that will make it 
> let some air out of the balloon when it finds things are a 
> tight fit. So effectively this should keep VMRM from choking 
> your penguins. But it means that remembering the setting will 
> not do you much good.

I have to admit it is a long time since I tried VMRM/CMM. A kernel fix
could certainly have been applied in the meantime. I did however find
VMRM/CMM rather simplistic - it is basically on/off per guest. There is
no way to control how much VMRM ask a Linux to shrink and I can't find
anything explaing how VMRM decides if VM has a memory constraint. I am
afraid it based of something similar to "emergency scan", which I
suspect most memory over commited VM does most of the time these days??

> 
> It's not too hard to have a server with secondary console set 
> up so you can display the current setting though.
> 
> When Linux is swapping, then in theory you should be able to 
> tell from the resident page frames and paged-out frames how 
> large the black hole is (because Linux does not grow). 
> Unfortunately, virtual machine pages can be both resident and 
> on paging DASD, so adding those up will not help you. There's 
> part of the virtual machine that does not cooperate in this 
> scheme, so you should be able to get some metrics by looking 
> at what happened earlier and how many pages were resident 
> before you paged it.

Still, a couple of CP monitor records for static and times page pools
would have been quite nice... 

Sincerly,
Klaus Johansen
__________________________________________________________________________________________
KMD A/S, Lautrupparken 40-42, DK-2750 Ballerup, CVR-nr. 26911745 

KMD er medlem af IT-Branchen og Dansk Erhverv samt anmeldt til Datatilsynet som 
edb-servicevirksomhed. KMD er certificeret i henhold til ISO 9001:2000, med 
Dansk Standard som certificerende organ og er desuden Microsoft Gold Certified 
Partner og Certificeret SAP Hosting Center.

www.kmd.dk   www.kundenet.kmd.dk   www.e-Boks.dk    www.organisator.dk   
www.kmdinternational.com

Hvis du har modtaget denne e-mail ved en fejl, bedes du venligst give mig 
besked herom og slette den.
If you received this e-mail by mistake, please notify me and delete it. Thank 
you.

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