On Thu, Nov 16, 2006 at 18:32:34 -0600, Laurence Vanek <lvanek@xxxxxxxxxxx> wrote:
> > Since its not if but when the next drive will fail, Im hoping I can > assemble a working plan. If not Im afraid I dont see the value in a > RAID1 setup.
I went through this about a year or so ago when I lost a drive from the mirror array (RAID-1) on my server. If I remember correctly, the sequence went something like this: 1) I got an e-mail from mdadm saying the drive had failed (plus the system was wonky but still somewhat functional). 2) I shutdown the system, pulled the ribbon cable from the failed drive and then powered back up. 3) mdadm detected that the drive was now missing but worked flawlessly in degraded mode. I ordered a replacement drive and waited. 4) New drive arrived so I plugged it in, partitioned it to match the drive I had pulled/remaining drive and then added each partition back into the array. 5) mdadm synced the new drive and all has been good ever since. I think the only data loss was during the period of time from when the drive failed until I was able to pull the ribbon cable. Unfortunately, the system was still trying to write to the dead drive. You'll need to dig through the man pages for mdadm for the specific incantations for the above steps. There may also be some variations if you are running LVM. I think the server was running CentOS 3 (RHEL 3 clone) at the time the drive died so no LVM. Cheers, Dave -- Politics, n. Strife of interests masquerading as a contest of principles. -- Ambrose Bierce
