page fault question

[Sonny_8]

Hi

I have monitor memory. I saw page fault
1600-2500 /second . Is there any problem on my
machine ,but page out is very low.
How to indicate memory problem.

Thanks
Sonny

[Chuck J]

Do a top or a glance and check if your mem usage is around 100% if so check what processes are using all the memory.

Chuck J

[Ravi_8]

Hi,

vmstat

in this look for page filed.
watch out pi and po.
if these two are very high, then you have to considering increasing swap or memory

[Chuck J]

Try #vmstat

and also here's a couple of threads:


http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x60865220af9bd5118ff10090279cd0f9,00.html
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x82e95f260cafd4118fef0090279cd0f9,00.html


Chuck J

[Eugeny Brychkov]

Sonny, memory statistics and deallocated space can be seen in PDC, or in hpux running diagnostics calles stm - cstm (command line), mstm (menued), xstm (graphical).
Usually people having 700/96 terminals use mstm. Run it, then select memory (at the end of the device list), go to menu, select Tools->Information->Run
Eugeny

[Sonny_8]

Hi all

After read the forum that chuck J give me.
seem to be pagein and page fault very high.
pagefault 1500-2000/S,
after use vmstat -z
the following the result afer 45 mins
# vmstat -s|grep page
82 pages swapped in
0 pages swapped out
1779693 page ins
936 page outs
117738 pages paged in
2741 pages paged out
1674065 total page reclaims
10 intransit blocking page faults
4382044 zero fill pages created
2146407 zero fill page faults
427246 executable fill pages created
14791 executable fill page faults
0 swap text pages found in free list
158688 inode text pages found in free list
48003 pages scanned for page out
2349 pages freed by the clock daemon

# swapinfo -t
Kb Kb Kb PCT START/ Kb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 4194304 682340 3511964 16% 0 - 1 /dev/vg00/lvol2
reserve - 1766452 -1766452
memory 1553220 565768 987452 36%
total 5747524 3014560 2732964 52% - 0 -
# swapinfo -t
Kb Kb Kb PCT START/ Kb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 4194304 682340 3511964 16% 0 - 1 /dev/vg00/lvol2
reserve - 1766452 -1766452
memory 1553220 565768 987452 36%
total 5747524 3014560 2732964 52% - 0 -
# swapinfo -t
Kb Kb Kb PCT START/ Kb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 4194304 682340 3511964 16% 0 - 1 /dev/vg00/lvol2
reserve - 1766452 -1766452
memory 1553220 565768 987452 36%
total 5747524 3014560 2732964 52% - 0 -


pls help.

Regards
Sonny

[Mike Stroyan]

Note that pageins and zero fills are completely normal when starting a new process. Each page of text or data from the program file is paged in the first time that a process accesses that page. If your system is continually running new processes it will be continually faulting in new pages. That is not necessarily an indication of a problem.

Some program could have a defect which caused it to use many more pages or start many more processes that it needed to get perform a task. In that contrived case the pageins would be part of a resource waste. However, tools like glance cannot tell you whether normal system activity is 'useful' activity.