Disk Utilization 100% always

[Amol Pathak]

On our HP UX 9000 machine, disk utilization is always 100%. From Glance, it is the disk from EVA, used for Oracle DB and engine. This is on cluster. How can we reduce the disk utilization? Any kernel parametr tuning required?

[Kapil Jha]

first you check if disk space available is enough for the data you have , because i suppose server with oracle are supposed to have huge data.
MOre info required....
BR,
Kapil

[Amol Pathak]

Have enough space. Here are filesystems.

/dev/vg01/data 25600000 22328992 3245456 87% /m01
/dev/vg01/index 15360000 11070400 4256096 72% /m02
/dev/vg01/rbsredo 10510336 7978096 2512464 76% /m03
/dev/vg01/arch 10240000 1051496 9116920 10% /m04
/dev/vg01/temp 5120000 2267416 2830304 44% /m05
/dev/vg01/export 30720000 8756392 21792096 29% /m06
/dev/vg01/engine12 5255168 4680536 570440 89% /oraclei


Also from Glance, the individual filesystem disk utilization.

9 /m01 /dev/vg01/data vxfs 587.1/587.1 1334/ 1334
10 /m02 /dev/vg01/index vxfs 14.2/ 14.2 14.2/ 14.2
11 /m03 /dev/vg01/rbsredo vxfs 5.7/ 5.7 5.7/ 5.7
12 /m04 /dev/vg01/arch vxfs 0.0/ 0.0 0.0/ 0.0
13 /m05 /dev/vg01/temp vxfs 138.5/138.5 178.5/178.5
14 /m06 /dev/vg01/export vxfs 0.0/ 0.0 0.0/ 0.0
15 /oraclei /dev/vg01/engine12 vxfs 112.8/112.8 1.4/ 1.4

[Kapil Jha]

first output is disk utilization so disks are awailable.
Second is IO.
Disk utilization is OK a bit high.
BR,
Kapil

[ssudhir]

check "iostat" or top to find out who is taking the huge memory.

also check for dmesg or /var/adm/syslog/syslog.log for any fiberchannel related errors .

check /var/opt/resmon/log/event.log for any errors.

[Srikanth Arunachalam]

Hi,

I think this should not be a cause of concern. The reason being EVA manages the IO distribution to the disk of itself and disk utilization cannot be attributed to user application or OS accessing the disk. It keeps monitoring the disk all the time. All you need to do is to monitor the statistics from the EVA StorageWorks Command View.

Thanks,
Srikanth

[Kapil Jha]

can you post vmstat
Kapil

[yulianto piyut]

tell to your DBA team to distribute the datafile. reconfgiure your lvm configuration to distribute the fc that you used (for primary & alternate)


-yut-