topic Re: find: cannot stat in Operating System - HP-UX

find: cannot stat

itrc55 — Mon, 17 Oct 2011 12:47:25 GMT

Hi,

when i try to execute this command

# find /comptel2/elink/data ! -user data -print

i facing this error

find: cannot stat /com/data/30/COLLECTED_0_1698/1-112/1318656117x001_0011698x_1
find: cannot stat /com/data/data/30/COLLECTED_0_1698/1-111

Although this file is available " 1318656117x001_0011698x_1 "

/com/data/30/COLLECTED_0_1698/1-112> ls

1318656117x001_0011698x_1

and this error message

Oct 17 15:21:27 ruhdbvmunix: vxfs: WARNING: msgcnt 819966 mesg 008: V-2-8: vx_direrr: vx_dirlook_4 - /com/data/30 file system dir inode 79180 dev/block 0/38715732 dirent inode 79513 error 6

and this is a file system, i using (VCS)

/dev/vx/dsk/medoffdg/datavol 1048576000 155273803 837542027 16% /com/data/30

pleas how can i resolve this issue ?

Thanks

Re: find: cannot stat

James R. Ferguson — Mon, 17 Oct 2011 13:04:42 GMT

Hi:

It might be helpful to tell us the operating system release and particularly the version of VxFS software you are running. That said, it appears that you may have corruption of the filesystem. Unmount the filesystem and perform a full 'fsck'.

Regards!

...JRF...

Re: find: cannot stat

Matti_Kurkela — Mon, 17 Oct 2011 13:14:19 GMT

> Oct 17 15:21:27 ruhdbvmunix: vxfs: WARNING: msgcnt 819966 mesg 008: V-2-8: vx_direrr: vx_dirlook_4 - /com/data/30 file system dir inode 79180 dev/block 0/38715732 dirent inode 79513 error 6

The filesystem may be corrupted (damaged). "cannot stat" means the filesystem cannot report all the fstat()/stat() information on the file, so something must be wrong with the file metadata. If you look at the problem files with "ll", do the file permissions/ownership/size etc. look sensible?

You should schedule some downtime, and then unmount the filesystem and run a full filesystem check.

You mentioned VCS (Veritas Cluster Suite?): if this filesystem is visible to multiple cluster nodes simultaneously, you'll need to unmount it from all of them before running a filesystem check.

If Serviceguard is involved, the cluster filesystem should unmount when you halt the appropriate multi-node packages. With VCS, you may have to do something similar: I'm not too familiar with VCS.

Once the filesystem is safely unmounted, a full filesystem check can be started like this:

fsck -F vxfs -o full /dev/vx/dsk/medoffdg/datavol

As this is a terabyte-sized filesystem, a full check for it may take a while.

Re: find: cannot stat

itrc55 — Tue, 18 Oct 2011 04:59:27 GMT

Hi,

the version of VxFS software

# fstyp -v /dev/vx/dsk/medoffdg/datavol
vxfs
version: 7
f_bsize: 8192
f_frsize: 1024
f_blocks: 1048576000
f_bfree: 892765077
f_bavail: 836967260
f_files: 223574756
f_ffree: 223191268
f_favail: 223191268
f_fsid: 33568434
f_basetype: vxfs
f_namemax: 254
f_magic: a501fcf5
f_featurebits: 0
f_flag: 16
f_fsindex: 9
f_size: 1048576000

# swlist | grep VCS
VRTSvcsvr 5.0.31.5 VERITAS Volume Replicator Agent for VCS

The operating system release is 11.31

And how can i check if this filesystem is visible to multiple cluster nodes simultaneously ? becuase also I'm not too familiar with VCS.

Thanks