topic Re: disk que in Operating System - HP-UX

disk que

pa8700 — Wed, 20 Jan 2010 18:54:54 GMT

device %busy avque r+w/s blks/s vwait avserv
Average c2t0d0 17.01 8.13 36 853 19.63 12.55
Average c0t0d0 24.51 6.78 44 952 17.45 14.42

See pls find the avg of the sar -d out put of one day..

My server is havig some performance issue and looking into it.

It seems the avserv is almost 14 and the avque is almost 8 .

This is for the VG00 disks. Not the applcication VG disks.

Do you see any bottle neck on the local disks...

Re: disk que

Suraj K Sankari — Wed, 20 Jan 2010 19:03:45 GMT

Hi,
For performance issue you need to check so many things some of are

Check what are the application is running?
how much physical memory your server have?
which process is taking more memory?
how much swap memory you have?

Suraj

Re: disk que

pa8700 — Wed, 20 Jan 2010 19:31:12 GMT

guys this is a SD64k withy 150 GB Mem ,no swaping....hapening 85% mem use..

seems some disk bottle neck..

Re: disk que

Fabian Briseño — Wed, 20 Jan 2010 20:06:10 GMT

seeing this.

%busy seems ok.

check command top

and vmstat -n

post results here.

The important question would be, what kind of performance issue. What application ?
is the whole system affected ?
since when is hte system with this issue ?

Re: disk que

pa8700 — Wed, 20 Jan 2010 21:10:53 GMT

$ vmstat -n
VM
memory page faults
avm free re at pi po fr de sr in sy cs
19778984 5713012 1042 230 13 0 0 0 0 51441 767348 30307
CPU
cpu procs
us sy id r b w
23 18 59 31 45 0
25 14 61
25 14 60
26 15 60
33 14 53
32 14 54
32 14 54
31 17 51
31 19 51
32 17 50
33 17 50
35 16 49
32 15 53
30 15 55
31 15 54
31 15 54
29 15 55
29 16 55
5 3 93
30 16 55
28 16 56
29 16 55
4 3 93
29 16 55
27 17 56
28 16 56
4 2 94
29 16 55
29 17 55
29 17 54
4 3 93
28 19 53

It seems falts are there more

Faults
in sy cs
51441 767348 30307

Re: disk que

pa8700 — Wed, 20 Jan 2010 21:11:59 GMT

again

what is this pac-cli process..

CPU TTY PID USERNAME PRI NI SIZE RES STATE TIME %WCPU %CPU COMMAND
29 ? 19477 root 241 20 36996K 15832K run 8028:36 95.67 95.50 pac-cli

Re: disk que

Fabian Briseño — Wed, 20 Jan 2010 21:18:20 GMT

Hello.
execute the following command:

top

let it run for a while

post results.

Re: disk que

Fabian Briseño — Wed, 20 Jan 2010 21:22:02 GMT

If I remember correctly pac-cli is a gathering tool used by HP.

it is used to anilze your system in our case I think it was gathering info on our SAN performance.

Re: disk que

Tim Nelson — Wed, 20 Jan 2010 22:01:52 GMT

That now explains your first performance issue..

the performance metric gatherer is causing the performance problem..

kill it.. ;)

Re: disk que

Kapil Jha — Thu, 21 Jan 2010 01:20:26 GMT

Can you check where is how much data pac-cli is dumping, if thats huge then its smthing to check out.

try to stop this process and see the performance.

Is it running everytime for once in a while.

BR,
Kapil+

Re: disk que

pa8700 — Thu, 21 Jan 2010 16:31:30 GMT

hi the pac-cli process killed.still its not improved...
in glance memory report it shows page faults....what can be the cause...
MEMORY REPORT Users= 3
Event Current Cumulative Current Rate Cum Rate High Rate
--------------------------------------------------------------------------------
Page Faults 37094 304052 7727.9 4081.2 8061.4
Page In 0 30 0.0 0.4 4058.6