IO Issues

[shruti_1]

one of my disc c18t5d0 is going approx 100% busy from two days. sar -d output is as follow. How do i find whether there is any issues with this disc. Pls see the

[Steven E. Protter]

Shalom,

You can check basic disk functionality with cstm mstm or xstm exercize function.

The problem though is most likely because you have too much activity on the disk in question. Its a sysadmin, identify the problem and configure your way out of the problem situation.

I'm guessing there is an active database or something similar on the disk. Maybe write intensive.

SEP

[Mustafa Gulercan]

Also you can check disk utility by using glance plus.
#glance

then press (d key) at glance menu.it shows local and remote I/O activity.The I/O distribution can be viewed from the following:
Logical reads/writes,physical reads/writes.

The glance disk device report (u key) shows current and average utilization of each disk drive on the system.

(v key) shows disk activity on a perl logical volume basis.Physical I/O activity is shown with this logical volume report.

[Albert_31]

Hello Shruti,

A disk being 100% does not imply any issue, it only informs that the disk is being accessed most of the time...and in your case, more than all the other disks, check what application is mounted on the VG which contains teh disk and see if there is any activity being carried out which has increased teh disk access..

regards

[Raj D.]

Hi Shruti ,

If you like to check the disk (c18t5d0 ) , you can check it with cstm.

1. Get the hardware path of the disk c18t5d0 , with # ioscan -fnC disk

2. # echo "sel path 0/1/... ;info;wait;infolog" | /usr/sbin/cstm

And check for Read error , Write error.

[ Where 0/1/... is the hardware path , it can be obtain from ioscan output. ]


Also check with diskinfo.


Cheers,
Raj.

[Alzhy]

Your sar stats for said disk appear perfectly healthy to me. Do you actually have any performance issues to talk about or is your concern just focused on this disk's being 100% busy?

The most important statistic to look for in "sar -d" output to ascertain if you've disk performance issues is the "avque" or disk queue field. Any value above 1.0 most often indicates the disk can no longer handle I/O requests.

Shalom.