If you are running build 36 or later, you can run:
# fmdump -eV
To get a verbose list of exactly what errors were seen. Note that we
don''t currently do anything with these ereports - the next phase of
ZFS/FMA will do some predictive analysis of these errors and proactively
fault questionable drives.
>From build 36 onwards, your app should get an immediate EIO when trying
to read the file. Exactly how tar behaves in this scenario I don''t
know.
- Eric
On Mon, Mar 27, 2006 at 06:19:43PM -0800, Darren Reed
wrote:> In the process of tar''ing up files in an older ZFS partition
> (23.6.2005), the tar command seized up. Truss showed it hanging
> in stat64(), so I went looking for symptoms.
>
> In "zpool status -ve", I found "4" in the SUM column.
>
> Being from the old school, I did "dmesg", expecting to see some
> kernel error message about the disk but found nothing.
>
> Is there something I can do to turn on reporting of disk errors
> as they occur?
>
> Of course I ran into the problem of not being able to get past
> the file causing the error but I suspect that''s a bug that has
> been fixed?
>
> FWIW, the problem of dealing with disk errors has always been
> why I''ve preferred to use ufsdump/ufsrestore than other tools
> for backups.
>
> Darren
>
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
--
Eric Schrock, Solaris Kernel Development http://blogs.sun.com/eschrock