Lot of pagefaults

[Sandip Ghosh]

Server - HP K460 2CPU 1.75GB
O/S - HP-UX 10.20

Whenever I am looking through the memory reports by Glance I found that there are lot of pagefaults. I have checked through mstm, it has not logged any error for the memory. Can anybody please let me know what are the possible area I can look for the page faults. Here is a output of glance memory report.

Page Faults 5 3402253 0.9 281.1 3663.0
Paging Requests 0 1252338 0.0 103.5 1403.0
KB Paged In 0kb 10.8mb 0.0 0.9 806596.1
KB Paged Out 0kb 4kb 0.0 0.0 0.7
Reactivations 0 0 0.0 0.0 0.0
Deactivations 0 0 0.0 0.0 0.0
KB Reactivated 0kb 0kb 0.0 0.0 0.0
KB Deactivated 0kb 4kb 0.0 0.0 0.7
VM Reads 0 642 0.0 0.0 24.0
VM Writes 0 0 0.0 0.0 0.0

Total VM : 835.7mb Sys Mem : 161.9mb User Mem: 1.25gb Phys Mem: 1.75gb
Active VM: 171.9mb Buf Cache: 268.8mb Free Mem: 77.9mb

[Peter Kloetgen]

Hi Sandip,

seems like your swapspace is not enough. Try to add some swapspace, as much, that you get at least as total RAM amount. Hope this will do it for you.

Allways stay on the bright side of life!

Peter

[Ian Dennison_1 · Accepted solution]

Page Faults are an inevitable part of managing memory.

They are not really a "fault", just an indication that the page reference could not be found in in the TLB Cache, so had to be looked for elsewhere. This is part opf the PA-RISC architecture.

I have tried to look for a resolution to this one myself, but have yet to find an answer other than "upgrade to a bigger CPU with a bigger cache".

(Most of this info was pilfered from the Memory Management White Paper)

Cheers, Ian

[harry d brown jr]

page fault

An event recorded when a process tries to execute code instructions or to reference a data page not resident in a process' mapped physical memory. The system must page-in the missing code or data to allow execution to continue.

Just a "feature" of virtual memory.

http://docs.hp.com/hpux/onlinedocs/5965-4641/5965-4641_top.html


live free or die
harry

[Sandip Ghosh]

Hi Peter,

There is no swapping at all in the swapdevices. It is swapping 0% on the swap devices.

/dev/vg00/lvol2 device 512 0 1
/dev/vg01/lvol9 device 1024 0 0
pseudo-swap memory 1329 396 na

Sandip

[A. Clay Stephenson]

Hi:

Page faults are perfectly normal and do not usually indicate a problem. It simply means that an address could not be resolved in the Translation Lookaside Buffer (TLB) and that the system then had to go out to the much slower memory. Remember the TLB is quite small and thus page faults are inevitable. Something that you should worry about is a high pageout rate; that indicates swapping (well, paging actually) and your performance will degrade.

Regards, Clay

[Roger Baptiste]

hi,

The value to look for is pageout (po) through the vmstat -n command. If it is showing numbers corresponding to po constantly then there is a memory pressure. It should be ideally 0.

HTH
raj