Over the past month we’ve picked up a couple of AdvFS read errors like the
following in /var/adm/messages:
Jun 17 21:10:58 venus vmunix: AdvFS I/O error:
Jun 17 21:10:58 venus vmunix: Domain#Fileset: stripe1_s_d#stripe1_s_f
Jun 17 21:10:58 venus vmunix: Mounted on: /stripe1
Jun 17 21:10:58 venus vmunix: Volume: /dev/disk/dsk29b
Jun 17 21:10:58 venus vmunix: Tag: 0x0000cf9a.80f7
Jun 17 21:10:58 venus vmunix: Page: 2811
Jun 17 21:10:58 venus vmunix: Block: 89619040
Jun 17 21:10:58 venus vmunix: Block count: 144
Jun 17 21:10:58 venus vmunix: Type of operation: Read
Jun 17 21:10:58 venus vmunix: Error: 5
Jun 17 21:10:58 venus vmunix: EEI: 0x6400
Jun 17 21:10:58 venus vmunix: I/O error appears to be due to a hardware
problem.
Jun 17 21:10:58 venus vmunix: Check the binary error log for details.
Jun 17 21:10:58 venus vmunix: To obtain the name of the file on which
Jun 17 21:10:58 venus vmunix: the error occurred, type the command:
Jun 17 21:10:58 venus vmunix: /sbin/advfs/tag2name /stripe1/.tags/53146
This mount point is a RAID5 on SAN (HSG80).
Are these types of errors something that require corrective action? I was
thinking of a full backup, reinitialize the RAID set, and then recreate the
AdvFS domain. Also, are these errors correctable in the sense that
eventually the O/S (AdvFS) will retry the READ and eventually succeed?
Much thanks,
Chris
Received on Wed Jul 20 2005 - 16:42:21 NZST