| 1 |
To all who replied to my distress signal, |
| 2 |
|
| 3 |
The repair turned out to be pretty painless. In two ways: |
| 4 |
|
| 5 |
First, getting quality advice from all of you sans the roasting I deserved |
| 6 |
;), and |
| 7 |
|
| 8 |
Second, gdisk fixed the gpt header and partition table easily (details |
| 9 |
below). After that, I rebooted, zfs recognized the disk, and then it |
| 10 |
started a resilver automatically. It was done a few minutes later, and now |
| 11 |
everything’s back to normal. |
| 12 |
|
| 13 |
Gdisk noted that both the main gpt header and main partition table were |
| 14 |
damaged, but the backups were ok. I bypassed gdisk’s offer to use either |
| 15 |
the current gpt or to create a blank gpt, because I didn’t understand |
| 16 |
exactly what “current” or “blank” meant. |
| 17 |
|
| 18 |
Instead, I invoked the recovery & transformation menu with “r”. Then I |
| 19 |
used “b” to rebuild the damaged main gpt header with the good backup, |
| 20 |
followed by “c” to restore the partition table from the good backup. I |
| 21 |
then printed the partition table. It looked exactly like the partition |
| 22 |
tables on the other disks of the same make and model in the zfs pool |
| 23 |
(modulo what looked like a unique zfs partition name). That made me |
| 24 |
comfortable, so I wrote the changes to disk, rebooted, and found everything |
| 25 |
back to normal after the resilver. |
| 26 |
|
| 27 |
Appreciate all the help. Thanks! |
| 28 |
|
| 29 |
John |
Archives