different issue on each disk in vg00

[stephen peng]

guys,
I met cute issues on a two-disk vg00 before.
startup messages says:
vxfs fsck: /dev/vg00/lvol8:file system had I/O error(s) on meta-data.
vxfs fsck: /dev/vg00/lvol8:file system had I/O error(s) on user data.
/dev/vg00/lvol8:log replay in progress
/dev/vg00/lvol8:log replay failed to clean file system
/dev/vg00/lvol8:file system is not clean, full fsck required
/dev/vg00/lvol8:pass0 - checking structural files
/dev/vg00/lvol8:pass1 - checking inode sanity and blocks
vxfs fsck: /dev/vg00/lvol8:fsck read failure bno = 260376, off = 0, len = 65536
/dev/vg00/lvol8:file system check failure, aborting ...
Couldn't fix vxfs filesystems, these filesystems will not be mounted.
The boot process continues, but please check and fix these after login.
lvdisplay /dev/vg00/lvol8 told:
LV Status available/stale
00063 /dev/dsk/c3t0d0s2 00912 stale /dev/dsk/c3t1d0s2 01056 current
00070 /dev/dsk/c3t0d0s2 00919 stale /dev/dsk/c3t1d0s2 01063 current
from lvdisplay, c3t0d0 should be failing, but when I replace c3t0d0 and resynced vg00, lvdisplay output remained the same! and the failure of filesystem /dev/vg00/lvol8's fsck showed c3t1d0 got problem too. maybe c3t1d0 cause this vg00 problem.
how to understand this issue? would a "current" PE turn out to be "failed" one?

regards
Stephen

[madhuchakkaravarthy]

hi

any incident happened?????

can u tell us what lvol8 mounted to.

try to do

dd if=/dev/vg00/lvol8 of=/dev/null bs=1024k

try to read blocks

do u have a mirror copy ????

regards

MC

[James R. Ferguson]

Hi Stephen:

> would a "current" PE turn out to be "failed" one?

This is possible. Seeing stale extents on a particular physical volume does not necessarily mean that that physical volume is the fail(ed|ing) one. See:

http://bizsupport.austin.hp.com/bc/docs/support/SupportManual/c01911837/c01911837.pdf

Regards!

...JRF...

[stephen peng]

OK,now I need to know what exactly "stale" stands for. from my case, the "stale" status looked like "copyed" from a â claimedâ good diskã

regards
Stephen

[Bill Hassell]

> vxfs fsck: /dev/vg00/lvol8:file system had I/O error(s) on user data.
> vxfs fsck: /dev/vg00/lvol8:fsck read failure bno = 260376, off = 0, len = 65536

Easy answer: you have a bad disk. Check syslog.log for diagnostic messages or run dd to read the disk. fsck CANNOT fix I/O errors.

> OK,now I need to know what exactly "stale" stands for.

Stale means that the extent does not match the primary disk. Since I/O errors exist, this is to be expected since a write operation has failed when the mirrored disk was being updated.

[stephen peng]

I felt interesting about this case, how could this happen to these two disks? was that filesystem( it was /var) corrupted first or disk failed first? any one get some most reasonable and possible explaination?

regards
Stephen

[Ismail Azad]

Hi stephen,

From your output it suggests that your lvol8 /var as you say has failed in phase2 of fsck. This could mean that a subdirectory is not linked to the parent one or one of the files in the filesystem is not allocated to an inode. That is why one of the symptoms of a corrupted filesystem is that a particular file could change places in the filesystem heirarchy. You have filesystem corruption which has eventually led to your problems.

Regards
Ismail Azad