Phobos
Dabbler
- Joined
- Sep 8, 2014
- Messages
- 25
Hi all,
I have a fairly standard system: X10-SLH-F, E3-1231 v3, 32 GB EEC RAM, 6x4TB WD Red. I have a mirrored boot pool with 2x16GB SanDisk Cruzer Fits (USB 2.0).
Upon updating to 9.10.1-U4, the system came up, but the boot pool was degraded, as one of the boot drives did not attach. I shutdown, reseated the drives, and brought the system back up. Now the pool status listed a single CKSUM error for one of the drives.
So, I decided to isolate the bad drive: I shut down, removed drive B, and booted up. No issue, so I assumed drive B had been corrupted. I figured I'd insert drive B, and recreate the mirror on the drive (using the Replace button on the boot pool status page). But, it immediately rejoined the pool when I inserted it.
Confused, I repeated this again with the drives switched. No errors. I finally relented and just let the system go along with what it claims is an okay pool. I ran a scrub and it found no errors.
So, what happened? How does the mirrored pool still work if the drives got out of “sync”? Wouldn't each drive have received a different set of writes? Or were there no writes because my system dataset (+syslog +rrd) is on my storage pool? Can I trust in zpool status and just assume that everything is okay?
In any case, I'm probably going to replace both USB drives so they don't fail at an inopportune moment, but I'm wondering if I can just mirror the existing pool onto the new drives, or if I should start from scratch.
Any advice/thoughts greatly appreciated.
I have a fairly standard system: X10-SLH-F, E3-1231 v3, 32 GB EEC RAM, 6x4TB WD Red. I have a mirrored boot pool with 2x16GB SanDisk Cruzer Fits (USB 2.0).
Upon updating to 9.10.1-U4, the system came up, but the boot pool was degraded, as one of the boot drives did not attach. I shutdown, reseated the drives, and brought the system back up. Now the pool status listed a single CKSUM error for one of the drives.
So, I decided to isolate the bad drive: I shut down, removed drive B, and booted up. No issue, so I assumed drive B had been corrupted. I figured I'd insert drive B, and recreate the mirror on the drive (using the Replace button on the boot pool status page). But, it immediately rejoined the pool when I inserted it.
Confused, I repeated this again with the drives switched. No errors. I finally relented and just let the system go along with what it claims is an okay pool. I ran a scrub and it found no errors.
So, what happened? How does the mirrored pool still work if the drives got out of “sync”? Wouldn't each drive have received a different set of writes? Or were there no writes because my system dataset (+syslog +rrd) is on my storage pool? Can I trust in zpool status and just assume that everything is okay?
In any case, I'm probably going to replace both USB drives so they don't fail at an inopportune moment, but I'm wondering if I can just mirror the existing pool onto the new drives, or if I should start from scratch.
Any advice/thoughts greatly appreciated.