ZFS can Panic When Storage Is Re-added (Doc ID 2070645.1)

Last updated on JANUARY 31, 2019

Applies to:

Symptoms

 Typical steps that can lead to this issue are:

1. Power off the old upper JBOD enclosure and remove it from the rack.
- Half of the mirrored pairs went "UNAVAIL" and the zpool went "DEGRADED", as expected.

2. Install new upper JBOD enclosure into the rack in the old upper enclosure's place, and power it on.

3. One by one, remove the physical disks from the old upper JBOD enclosure and insert them to the new JBOD enclosure (this requires a tray swap, which takes a few minutes per disk).
- After a minute or two, most of the disks came "ONLINE" on their own, but for any that didn't come online by themselves we ran "fmadm repaired <disk_id>" and they came online.

4. Once all the disks are swapped, check zpool status and confirm/wait until all resilvering are complete.
- The resilvering were all done by the time we got back to the console and ran the zpool status command.

5. Power off the old lower JBOD enclosure and remove it from the rack.
- Half of the mirrored pairs went "UNAVAIL" and the zpool went "DEGRADED", as expected.

6. Install new lower JBOD enclosure into the rack in the old lower enclosure's place, and power it on.

7. One by one, remove the physical disks from the old lower JBOD enclosure and insert them into the new JBOD enclosure (this requires a tray swap, which takes a few minutes per disk).

==== Panic occurred after the 2nd or 3rd disk was inserted in the lower JBOD enclosure ===

From this point onward, the zpool can be imported only after the second tray is physically powered off (and one can detach the "UNAVAIL" vdevs to get get non-mirrored zpool as "healthy" configuration). But, every attempt to attach the disk from that JBOD will result in panic similar to:

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.