I have a 4x500 SATA RAIDZ and one drive shows signs of failing (one bad sector can't be relocated using vendor utility or badblocks). Zpool scrub shows the data is OK but I get constant smartd warnings. I have other reasons to believe this drive is genuinely sick. I have ordered a new replacement drive but I'm not sure how to proceed, having failed to accomplish this once before. I believe the previous failure was due to partitioning problems (replacement drive already had MBR partitions on it from a previous application).
The issue is complicated by the fact that if I add in the new drive it will have to go on a separate SATA controller, and when I remove the failing drive, I will move it back to the same port the failing drive is currently on, and remove the temporary controller, so the drive name will change.
My assumptions:
1. It's better to put in the replacement drive on the temporary SATA port and do the zpool replace with all the current data available (since the old drive hasn't failed yet) instead of pulling the old drive and doing the replace with only 3 out of 4 drives. I read this somewhere.
2. I will have problems when I relocate the new drive to a different port and it assumes a different name.
3. I need to partition the replacement drive in advance using gpart, creating a 2GB swap partition and a BSD ZFS partition for the rest of the drive.
If anyone could confirm or deny these assumptions, and tell me how to handle the drive name change if I proceed as I described, I would appreciate it. Or if it's safer just to pull the old drive and then replace, and it will just take longer to rebuild, I would do that.
I'm also willing to wait for 8.0.1, which seems to have fixed some of the zpool replace issues, I'm just not convinced that it will handle the temporary port problem.
The issue is complicated by the fact that if I add in the new drive it will have to go on a separate SATA controller, and when I remove the failing drive, I will move it back to the same port the failing drive is currently on, and remove the temporary controller, so the drive name will change.
My assumptions:
1. It's better to put in the replacement drive on the temporary SATA port and do the zpool replace with all the current data available (since the old drive hasn't failed yet) instead of pulling the old drive and doing the replace with only 3 out of 4 drives. I read this somewhere.
2. I will have problems when I relocate the new drive to a different port and it assumes a different name.
3. I need to partition the replacement drive in advance using gpart, creating a 2GB swap partition and a BSD ZFS partition for the rest of the drive.
If anyone could confirm or deny these assumptions, and tell me how to handle the drive name change if I proceed as I described, I would appreciate it. Or if it's safer just to pull the old drive and then replace, and it will just take longer to rebuild, I would do that.
I'm also willing to wait for 8.0.1, which seems to have fixed some of the zpool replace issues, I'm just not convinced that it will handle the temporary port problem.
