Oracle App getting the following error when trying to write to a mountpoint

[KPS]

oracle > oerr ora 1114
01114, 00000, "IO error writing block to file %s (block # %s)"
// *Cause: Device on which the file resides is probably offline
// *Action: Restore access to the device


I have checked the following to makesure the integrity of the Logical Volume is good"

- lvdisplay -v /dev/vg01/lvol1 | more
and looked for stale extents

- checked the syslog
- checked dmesg
- did a pvdisplay -v /dev/dsk/c#t#d# to check for stale extents and to see if the system sees the disk it's writing to as unavailable.
All looks fine
- the disk we're talking about here is external EMC Disk in a Symmetrix.


Has anyone ever seen this type of error for those that are running Oracle Apps and/or can you think of any other ways to troubleshoot this???

Tks,
-Ken

[bhoopathi]

Hi Ken

Can you check if ioscan shows whether the disk is claimed or not. Do

ioscan -fnC disk

and check if the status of the disk shows claimed.

Thanks

[KPS]

That's one thing I didn't think to check and all looks good with an ioscan on the disk.

The disk devices all show claimed and specifically the devices in question.

[S.K. Chan]

This may be Oracle related if you cannot find anything wrong with the device/disk. Possibilities may include ..
- The file it is trying to write to is not accessible (permission perhaps or some other reasons).
- The file is corrupted.
Just some thoughts..

[Pradeep_3 · Accepted solution]

Some times pv timeout, which is not set properly causes this kind of problem, but generally it reports in syslog.log and dmesg. But any way you can check and change pv timeout value to recommended size i.e. 180 for EMC disks.

You can do this task without disturbing your current or running environment online by giving following command

pvchange -t 180 /dev/dsk/cXtYdZ

You can change this value for all the disks even if they are not giving any problem at this time as recommended by HP.

Once you try this revert back with the result on your problem, If it still persist further can be checked.

Paresh