[OpenIndiana-discuss] RaidZ-2 Array Failure - Trying to recover

[Jan Owoc]

Hi EchoBinary,


On Sat, Feb 25, 2012 at 2:09 AM, Echo Binary <echobinary at gmail.com> wrote:
> So okay - I have a controller card with 12 internal SATAII ports and
> have been successfully running an array of all 1.5TB disks for the
> past few years. First on OpenSolaris, then on OpenIndiana.
>
> A few months ago 1 disk failed and I was of course concerned, but
> being a RaidZ-2 I rolled the dice and hoped there would not be another
> failure.  Then around Christmas I got the funds to replace the disk -
> only to discover after getting back home that there was another disk
> failure.
>
> It is my understanding that the RaidZ-2 array can remain functional
> with 1 failed disk. But with 2 failed - It will not function but can
> still be recovered - yes?

RAID-Z2 should be able to sustain a complete failure of two drives
with no data loss, and remain online. However, in a double-degraded
mode, any further read errors in the data would result in data loss.

I understand that last you were able to run "zpool status", you had
two entirely failed drives that had been automatically removed from
the zpool, and no errors on any of the other drives.


> So - I have the 2 brand new replacement drives, i swapped them in, and
> tried the zfs replace command. etc...   However it keeps telling me
> that the array canot replace the drives because it is not online. I
> try to online it and the command locks up- forcing me to close that
> terminal window.

Could you post the most recent "zpool status" you were able to run
(i.e. how do you know which drive was failed), as well as the error
message you are getting when you try to replace? (zpool status can
sometimes take several seconds to finish, especially on degraded
pools)


> From then on zpool status and zfs list just lock up the command line and hang.

I'm not sure if this is the best solution, but I would try to zpool
export / zpool import, or otherwise try to refresh the configuration
(reboot?) of the pool. The risk with the export/import or a reboot is
that if something else breaks that you can't import the pool, you
could lose the data entirely.


> Any pointers or help would be great!

I think this documentation is most relevant to your problem:
http://docs.oracle.com/cd/E19120-01/open.solaris/817-2271/gbctt/index.html


Jan