[zfs-discuss] Viewing repaired files

SanMadJack sanmadjack at gmail.com
Thu Sep 19 09:01:28 EDT 2013

That's good advice, though a little cost-prohibitive for my budget. :D
Currently my setup looks like this:

        NAME                                     STATE
        tank                                     ONLINE
          mirror-0                               ONLINE
            scsi-SATA_TOSHIBA_DT01ACA_Y2Q4V47AS  ONLINE   (Primary
            scsi-SATA_ST2000DM001-9YN_S2F0AJV7   ONLINE       (Secondary
          mirror-1                               ONLINE
            scsi-SATA_ST2000DM001-9YN_Z240CGCE   ONLINE      (Primary
            scsi-SATA_ST2000DM001-9YN_Z2F0FFJC   ONLINE       (Secondary

So if either controller catastrophically fails, I'll only lose one of the
drives from each mirrored pair, leaving me with two drives that I should be
able to use to get uncorrupted data from.

^Matt(hew)?[ ]Barbour$

On Thu, Sep 19, 2013 at 5:45 AM, Gregor Kopka <gregor at kopka.net> wrote:

> Am 18.09.2013 23:38, schrieb SanMadJack:
>  Absolutely; a bad controller or (probably more likely) a drive that is
>> starting to fail. The first thing I would do in your situation would
>> be to check the SMART status of the problematic drive, paying
>> particular attention to the reallocated sector count. If it's more
>> than a small handful and especially if it increases over time,
>> strongly consider replacing the drive ASAP and while ZFS is still able
>> to deal with the errors without problems. The instant you see an error
>> in the same spot on both drives, you have data loss, and that
>> probability goes up very fast once either drive starts seeing bad
>> sectors.
>  I checked my SMART stats, they're all fine. I've got a 4 disc setup, two
> pairs of mirrors. The secondary disks in each mirror both had checksum
> errors, and those two disks are on the same controller. I'm going to wait a
> few days and then scrub again, see if the same results pop up. If they do,
> I'll try replacing the controller and see how it goes from there.
> I suppose while we're discussing it, have any advice for best disk
> distribution across SATA controllers?
> Attach the members of each mirror to different controllers, so in case one
> controller dies (or corrupts data) your mirror is able to recover from the
> other drive (in case both/all mirror members are on the dying controller
> the pool would be corrupted instantly and would need restore from backup).
> Gregor

To unsubscribe from this group and stop receiving emails from it, send an email to zfs-discuss+unsubscribe at zfsonlinux.org.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.zfsonlinux.org/pipermail/zfs-discuss/attachments/20130919/bebff470/attachment.html>

More information about the zfs-discuss mailing list