topic Re: EMS messages about disk media errors in Operating System - HP-UX

EMS messages about disk media errors

Dave Chamberlin — Mon, 03 Jun 2002 13:30:10 GMT

About 3 or 4 times a week, I am getting a Critical EMS notification on my most active disk. The details show that the system was unable to perform a read due to an error on the medium. Since the utilization on this disk is constantly high, I am wondering if the disk really needs replacing or if the read is failing under heavy load for some other reason. The disk is half of a mirror - the other disk in the mirror never shows errors. Can someone give advice here? Thanks.

Re: EMS messages about disk media errors

Sanjay_6 — Mon, 03 Jun 2002 13:34:28 GMT

Hi Dave,

If EMS is reporting a mediaum error, you should replace the disk. If it is possible, you can reduce the mirror on the disk, remove it from the VG and then do a READ/WRite (Destructive) exercise on this disk using STM. If STM says that the disk has medium error, replace the disk.

Hope this helps.

Regds

Re: EMS messages about disk media errors

Helen French — Mon, 03 Jun 2002 13:46:06 GMT

As suggested by Sanjay, you can make use of the STM commands here:

# stm

This will display each disks on the system and check the specific disks with stm tools. If you have any errors in the media, it will report it. If found any error, you need to replace it with a new one. You can also check the disk with dd command:

# dd if=/dev/dsk/cxtydz of=/dev/null bs=1024K

EMS errors normally indicates hardware errors.

Re: EMS messages about disk media errors

S.K. Chan — Mon, 03 Jun 2002 14:18:44 GMT

The sequence of processes that I would normally do whaen I get a suspicious EMS notification that indicate a potential bad disk. I usually start with the easy ones and if they are already indicating some hardware errors at the begining, your chance of a disk failure is higher.
1) Run /etc/dmesg and check /var/adm/syslog/syslog.log for any error.
2) Run "pvdisplay -v /dev/rdsk/cXtYdZ | more" to check for any stale extents.
3) Run "/etc/diskinfo /dev/rdsk/cXtYdZ" to see if disk response.
4) Run cstm info ..
# cstm
cstm> map
cstm> sel dev
cstm> info
cstm> infolog
5) Run cstm exerciser ..
cstm> map
cstm> sel dev
cstm> exc
cstm> einf
If 1,2 & 3 are showing the signs, don't bother running cstm, just get the disk replaced.
Hope this helps ..

Re: EMS messages about disk media errors

Dave Chamberlin — Mon, 03 Jun 2002 20:15:42 GMT

Thanks for the suggestions. As it turns out, that disk is under contract and HP is going to come out at look at it and replace if necessary.