zfs discuss - Dec 2009 - Force file read even with checksum error

Hi guys,
I have a pool made with three luns striped.
After some scsi retryable messages, happened during a storage activity, zpool
status start to report one checksum error on one file only.
The zpool scrub find it but don''t solve it, and when I try to read the
file, I get an I/O error.
Again, when I try to copy the file, I get only some bytes copied, and then
stops, I mean this is because the next block is in error...
However, I''d like to do some check on this file, especially since I am
not convinced that the lun is bad and the file corrupted...
Is there a way to force zfs to give me the whole file so I can compare it to my
backup?
Is there a way to get zfs to give more specifics on the checksum issue (i.e.
which block, where in file, etc.)?

Thanks,
Stefano


Stefano Pini
Senior Technical Specialist at SUN Microsystems Spa
Viale Fulvio Testi 327 20162 Milano Italy
Contact | stefano.pini at sun.com - www.sun.com/italy

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20091210/ccc58dc8/attachment.html>

On Dec 10, 2009, at 2:09 PM, Stefano Pini wrote:
> Hi guys,
> I have a pool made with three luns striped.
> After some scsi retryable messages, happened during a storage  
> activity, zpool status start to report one checksum error on one  
> file only.
> The zpool scrub find it but don''t solve it, and when I try to read
> the file, I get an I/O error.
> Again, when I try to copy the file, I get only some bytes copied,  
> and then stops, I mean this is because the next block is in error...
> However, I''d like to do some check on this file, especially since
I
> am not convinced that the lun is bad and the file corrupted...
> Is there a way to force zfs to give me the whole file so I can  
> compare it to my backup?
> Is there a way to get zfs to give more specifics on the checksum  
> issue (i.e. which block, where in file, etc.)?
I wrote a few blog entries on this sort of thing.
http://blogs.sun.com/relling/entry/holy_smokes_a_holey_file
  -- richard

Hi Stefano,

I think you are saying that you have a non-redundant configuration
with data striped across 3 LUNs that is experiencing a disk problem
as seen by the SCSI retry messages and the I/O error when accessing
this file. ZFS can''t provide a whole file in a non-redundant
configuration.

See Richard''s follow-up about possibly accessing the remaining file
contents.

In the meantime, I would find the root cause of the hardware issue
by reviewing the /var/adm/messages file and the fmdump -eV output
to determine the root cause of the disk problem.

Thanks,

Cindy

On 12/10/09 15:09, Stefano Pini wrote:
> 
> Hi guys,
> I have a pool made with three luns striped.
> After some scsi retryable messages, happened during a storage activity, 
> /zpool status/ start to report one checksum error on one file only.
> The /zpool scrub/ find it but don''t solve it, and when I try to
read the
> file, I get an I/O error. 
> Again, when I try to copy the file, I get only some bytes copied, and 
> then stops, I mean this is because the next block is in error...
> However, I''d like to do some check on this file, especially since
I am
> not convinced that the lun is bad and the file corrupted...
> Is there a way to force zfs to give me the whole file so I can compare 
> it to my backup?
> Is there a way to get zfs to give more specifics on the checksum issue 
> (i.e. which block, where in file, etc.)?
> 
> Thanks,
> Stefano
> 
> 
> Stefano Pini
> Senior Technical Specialist at SUN Microsystems Spa
> Viale Fulvio Testi 327 20162 Milano Italy
> Contact | stefano.pini at sun.com 
> <mailto:stefano.pini at sun.com> - www.sun.com/italy
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/zfs-discuss