0506-324 I/O errors on filesystems: corrupt jfslog
Env: 3.2.5, RAID 5 unit.
Desc: Had been receiving DISK_ERR2 errors, but not any more.
Trying to mount a filesystem gives the following error:
0506-324 cannot mount /dev/lv06 on /oracle/data:
There is an input or output error
The customer has run fsck, which corrected a dirty superblock.
She got I/O errors trying to extend a filesystem, but was
successful in doing so.
Action: The volume group is lraid02 with one disk, hdisk10. There were
DISK_ERR2 errors in the log, but all occurred yesterday.
\# lsvg -p lraid02 : hdisk10 active
Had this disk really failed, we would expect the volume group
to have been forcibly varied off, with hdisk10 missing. Since
this is a RAID unit, there may still be hardware problems with
one or more disks in the RAID unit, but this shouldn't affect the
data due to the distributed parity.
We unmounted the other filesystem in this volume group, and ran
"logform /dev/loglv10". We tried to mount /dev/lv06, and got
errors about a dirty superblock. We ran "fsck -y /dev/lv06"
again to fix that. When we tried to mount the filesystem, we
still got I/O errors.
Commands like "iostat" work fine on the drive. We tried running
"dd if=/dev/LVname of=/dev/null bs=4k" to see if any I/O errors
were returned; none were.
We tried mounting these filesystems as read-only, thereby avoiding
the log device ("mount -r /dev/LVname"); this was successful.
The corruption seems to be coming from the log device itself. We
removed it and recreated it, and now the filesystems would mount
with no errors.
Support Line: 0506-324 I/O errors on filesystems: corrupt jfslog ITEM: AJ3982L
Dated: May 1995 Category: N/A
This HTML file was generated 99/06/24~13:30:27
Comments or suggestions?