Originally posted by: syedsaleembasha

Dear Experts,

we are running oracle database on p570 machine with AIX 5.3. since few days the server is slow and when i run topas it shows the hdiskpower2 disk always 100%. i have attached the fmon report also. Can you please give me suggest to improve performance.

other details:

lsvg -l rootvg
rootvg:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
hd5 boot 1 1 1 closed/syncd N/A
hd6 paging 106 106 1 open/syncd N/A
hd8 jfs2log 1 1 1 open/syncd N/A
hd4 jfs2 56 56 1 open/syncd /
hd2 jfs2 24 24 1 open/syncd /usr
hd9var jfs2 16 16 1 open/syncd /var
hd3 jfs2 32 32 1 open/syncd /tmp
hd1 jfs2 8 8 1 open/syncd /home
hd10opt jfs2 16 16 1 open/syncd /opt
lg_dumplv sysdump 8 8 1 open/syncd N/A
crs jfs2 48 48 1 open/syncd /crs
loglv01 jfslog 1 1 1 closed/syncd N/A
fslv02 jfs2 320 320 1 open/syncd /oradsk
oracle jfs2 200 200 1 open/syncd /oracle

lsvg -l oraappvg
oraappvg:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
loglv00 jfs2log 1 1 1 open/syncd N/A
fslv00 jfs2 700 700 2 open/syncd /u02

lsvg -p rootvg
rootvg:
PV_NAME PV STATE TOTAL PPs FREE PPs FREE DISTRIBUTION
hdisk1 active 546 29 00..00..00..00..29
hdisk3 active 546 226 08..00..00..109..109

lsvg -p oraappvg
oraappvg:
PV_NAME PV STATE TOTAL PPs FREE PPs FREE DISTRIBUTION
hdiskpower2 active 534 0 00..00..00..00..00
hdiskpower6 active 267 100 00..00..00..46..54

regards
#AIX-Forum

Originally posted by: nagger

Well not much information to work on here.
Running a database on so few disks would normally be a very bad idea.
Database should be spread on lots of disks.
Your VG ideas are bad too - one VG would let you spread out I/O on more disks across the VG.

But you should really ask your EMC admin why the one disk is so slow!

ta Nigel
#AIX-Forum

Originally posted by: nicofr

With SAN disks, you can see only 1 disks on the system for example but under it is spread on many disks...

Like EMC disks are often RAID5, the vg config is good !
#AIX-Forum

Originally posted by: shargus

Start with powermt display to show PowerPath's summary. Look for degraded paths, etc.

Then, try "powermt display every=1" and watch the IO/Sec and Q-IOs columns. If you have a number of Q-IOs with no IO/Sec, there's a problem with that path.
I've seen cases where there's a problem between the fiber switch and the EMC frame - Powerpath sees a good path, but I/O gets stuck. Eventually, PP fails over to the other path (eventually).
#AIX-Forum

Originally posted by: nicofr

Maybe you have to check with your SAN administrator.
We have EMC SANs and each time disks were at used at 100%, it was a EMC problem !
#AIX-Forum

Originally posted by: SystemAdmin

Also something may have changed on the switch (or at least someone just accessed it). Whenever I've had problems with EMC storage the switch was always the culprit. The reason for this is Powerpath. It was orginally developed for Windows but EMC configured it to be used for Unix as well but it can be a little flakey if changes are made to the switch

Check your fibre card config as well. If you do a lsattr -El fscsix make sure you have the following entries:

dyntrk yes
fc_err_recov fast_fail
This will help avoid problems if changes are made to the switch.
#AIX-Forum