Originally posted by: SystemAdmin

Our shiny new AIX 5.3 box is running very slowly. One thing we've noticed is the number of page faults - of the order of 20,000 per second. Should we be worried about this?

(We've got 6GB of RAM allocated to the machine, and it's using pretty much no paging space).

Topas said this:

Topas Monitor for host: groupapp EVENTS/QUEUES FILE/TTY
Mon Oct 23 16:49:51 2006 Interval: 2 Cswitch 5945 Readch 4243.3K
Syscall 22185 Writech 152.6K
Kernel 49.1 |############## | Reads 1898 Rawin 0
User 40.3 |############ | Writes 752 Ttyout 2344
Wait 0.1 |# | Forks 6 Igets 0
Idle 10.6 |#### | Execs 8 Namei 3514
Physc = 2.67 %Entc= 127.1 Runqueue 3.0 Dirblk 0
Waitqueue 15.5
Network KBPS I-Pack O-Pack KB-In KB-Out
en0 379.1 658.0 668.5 243.9 135.3 PAGING MEMORY
lo0 125.4 214.0 214.0 62.7 62.7 Faults 24282 Real,MB 6208
en1 95.7 144.5 56.5 28.7 67.0 Steals 0 % Comp 74.1
en2 5.6 8.5 8.5 2.7 2.9 PgspIn 0 % Noncomp 25.3
PgspOut 0 % Client 26.0
Disk Busy% KBPS TPS KB-Read KB-Writ PageIn 0
dac0-utm 0.0 0.0 0.0 0.0 0.0 PageOut 0 PAGING SPACE
dac1 0.0 0.0 0.0 0.0 0.0 Sios 0 Size,MB 5120
dac0 0.0 0.0 0.0 0.0 0.0 % Used 0.9
dac1-utm 0.0 0.0 0.0 0.0 0.0 NFS (calls/sec) % Free 99.0
ServerV2 0
Name PID CPU% PgSp Owner ClientV2 0 Press:
java 1097848 11.2 3.2 root ServerV3 0 "h" for help
java 1814582 3.4 15.3 root ClientV3 0 "q" to quit
oradb10 712876 0.8 5.4 root
java 684138 0.4 153.5 root

The java process 1097848 is the main websphere process.

cheers,
clive
#AIX-Forum

Originally posted by: andycor

Check your paging space.

Waitqueue 15.5 ---> A lot of process in the wait queue usually are there because paging space issues.

¿Do you have enough paging space? ¿How is configured?

Regards.
#AIX-Forum

Originally posted by: SystemAdmin

We've got 6GB physical memory, and 5GB pagespace. Nmon reports that we're not using that pagespace at all.
Re how is it configured : How do I tell? (please excuse my ignorance!)
#AIX-Forum

Originally posted by: andycor

I mean is all the paging space in one disk or more? You only use hd6? What size is it? How is distributed in case you use more disks?
What's the output of "lsps -a"?
#AIX-Forum

Originally posted by: SystemAdmin

Thanks for putting up with my ignorance :-)

Yes, only one disk.

Page Space Physical Volume Volume Group Size %Used Active Auto Type
hd6 hdisk0 rootvg 5120MB 6 yes yes lv

(although this disk is part of a mirrored pair, shared with another lpar - I've just been told it's a DS4300 array)

cheers,
clive
#AIX-Forum

Originally posted by: andycor

Hi,

I think this is not related to paging space, because there are not PgspIn & PgspOut activity and the lsps -a output shows only 6% used (~300MB).

Is hard to find the problem with just a topas snapshot. I just can see that are more processes in the wait queue than in the run queue, and this is bad. But this could happend just at the moment of the snapshot, only you can evaluate if it happends all the time.
Also, maybe there are problems with the file caching (page fault), check the maxperm vmo parameter, reduce the value, and see if it improve.

Regards,
Andres Cordoba.
#AIX-Forum