help !!! /usr file system inode question

[廖琼城]

syslog.log:
Feb 7 05:51:15 stpdb vmunix: SCSI: Async write error -- dev: b 31 0x01f000, err no: 5, resid: 8192,
Feb 7 05:51:15 stpdb vmunix: blkno:3544792, sectno: 7089584, offset: -665100288, bcount: 8192.
Feb 7 05:51:15 stpdb vmunix: vxfs: mesg 055: vx_metaioerr - /dev/vg00/lvo7 file system meta data write error
Feb 7 05:51:15 stpdb vmunix: vxfs: mesg 017: vx_ilisterr - /usr file system inode 274 marked bad
Feb 7 05:51:15 stpdb vmunix: vxfs: mesg 017: vx_ilisterr - /usr file system error reading inode 274

These log meaning the hardware have problem ?

I have run fsck -F vxfs -o full /dev/vg00/rlvol7 ,remove the inode 247, reboot machine, the system and application is normal, it will have some problem?

[Steven E. Protter]

Shalom,

if the fsck indicates that the problem was solved, AND the message stops in /var/adm/syslog/syslog.log then you are fine.

If the message persists, you have a hardware problem and need to back up the data.

You should check your disk configuration and see what bad block relocation is set for. The system can move data off blocks that are flakey but in general, if I have flakey blocks, I have a new disk on the way from HP Hardware support.

SEP

[Arunvijai_4]

Hello,

Try to run,

dd if=/dev/rdsk/cXtYdZ of=/dev/null bs=64K count=1000

-Arun

[廖琼城]

thanks: the message haved stops in /var/adm/syslog/syslog.log, it meaning fine ?

[Devender Khatana]

Hi,

This indicates that there is some error on the disk. The same can be confirmed by doing a red test on the disk. Rather then doing it on the whole disk as mentioned in earlier post I would recommend it to be done only on the specified LVOL by

#dd if=/dev/vg00/lvol7 of=/dev/null bs=1024k

You are suspected to have some more errors reported when this is run. I would suggest to mirror disk if not allready done and take an ignite backup.

HTH,
Devender

[Devender Khatana]

Hi,

This does not mean that everything is fine. This only means that the bad part of disk is not accessed now. You can be sure by running a read test on the LVOL, and the command repeat without error and there are no further errors in syslog.log then you are fine.

HTH,
Devender

[廖琼城]

Devender Khatana:

thanks,use #dd if=/dev/vg00/lvol7 of=/dev/null bs=1024k command is read test the LVOL? i will try it.

[Devender Khatana]

Hi,

Yes it is a read test and will read every block of the device mantioned, here lvol7 i.e. /usr file system.

HTH,
Devender