Re: Bad drive discovered during raid5 reshape

[Kyle Stuart <kstuart@sisna.com>]

David Greaves wrote:
> I read that he aborted it, then removed both drives before giving md a chance to
> restart.
>
> He said:
> After several minutes dmesg indicated that mdadm gave up and
> the grow process stopped. After googling around I tried the solutions
> that seemed most likely to work, including removing the new drives with
> mdadm --remove --force /dev/md1 /dev/sd[bc]1 and rebooting
>
> and *then* he: "ran mdadm -Af /dev/md1."
>   
This is correct. I first removed sdb and sdc then rebooted and ran mdadm
-Af /dev/md1.
>
> Kyle - I think you need to clarify this as it may not be too bad. Apologies if I
> misread something and sdc is bad too :)
>
> It may be an idea to let us (Neil) know what you've done and if you've done any
> writes to any devices before trying this assemble.
>
> David
When I sent the first email I thought only sdb had failed. After digging
into the log files it appears sdc also reported several bad blocks
during the grow. This is what I get for not testing cheap refurbed
drives before trusting them with my data, but hindsight is 20/20.
Fortunately all of the important data is backed up so if I can't recover
anything using Neil's suggestions it's not a total loss.

Thank you both for the help.