Drive Replacement Procedure via Nexenta NMC

=======================================================================
 Drive Replacement Procedure (针对raidz ，mirror，syspool）方法各不相同
=======================================================================

Determine which slot should be replaced
First, an important note: using non-redundant device configuration within a ZFS volume is NOT recommended.
1. Identify faulted logical drives using the following NMC command: 'show volume VOLNAME status'.

Example:
  nmc$ show volume data status
  data DEGRADED           0 0 0
      raidz2 DEGRADED     0 0 0
          c1t0d0 ONLINE  0 0 0
          c1t1d0 FAULTED 17 1.25K 0 too many errors
          c1t2d0 ONLINE  0 0 0
          c1t3d0 ONLINE  0 0 0
In the example above the faulted drive is c1t1d0

2a. If you have vendor provided slot mapping, identify the slot number by executing 'show lun slotmap'.

Example:
  nmc$ show lun slotmap
      +-------------------------------------------------+
     /                                                 /|
    /                                                 / +
   +-------------------------------------------------+ /
   | [ slot 01 ] [ slot 02 ] [ slot 03 ] [ slot 04 ] |/
   +-------------------------------------------------+
              NexentaStor - Model z200
  LUN     Slot#  DeviceId
  c1t0d0  1      id1,sd@f0000000048e317490009ec3b0000
  c1t1d0  3      id1,sd@f0000000048e54f7d000c091a0000
  c1t2d0  4      id1,sd@f0000000048e54f7e00001d2f0001
  c1t3d0  2      id1,sd@f0000000048e54f7e000226490002

2b. If you do not have slot mapping pre-configured, you could still identify the faulted drive using the following command: 'show lun LUN blink'.

Example:
  nmc$ show lun c1t1d0 blink
  Enabled blinking LED activity for disk 'c1t1d0' (press Ctrl-C to
  interrupt)...  ^CInterrupted

3. Once physical location of the faulted drive is identified, replace the faulted drive with a new drive of the same or bigger size, and execute the following steps

--------------------------------------------------------------
Replace Raidz/Mirrored disk in the same slot:
---------------------------------------------------------------
This depends on whether your vdevs are made up of mirrors or raidz.
For mirrored vdevs, you can detach the failed disk using
nmc$setup volume 'vol_name' detach-lun
However, this will leave the pool without redundancy. In this case it is preferable to attach a 3rd disk to the mirror creating a 3 way mirror, before doing the detach-lun. If you do not have a spare slot, but have spares, you could temporarily remove a spare to free up a slot.
Once the detach-lun has been run, the disk can be removed from the slot and a new one inserted. This disk can then be added to the pool again using
nmc$setup volume 'vol_name' attach-lun
If you are usingRaidZ, disks cannot be detached. In this case you could offline the disk using
nmc$setup volume 'vol_name' offline-lun
At this stage the disk can be removed from the slot and a new one inserted. This can then be added to the pool using

Example:

  nmc$ setup volume data replace-lun -r -y
  LUN to replace              : c1t1d0
  Please confirm using the same 'c1t2d0' as a replacement? If unsure,
  say 'No' or 'n' and re-run the command with -h option  (y/n) Yes
The above will replace your newly inserted physical disk in the volume 'data'. In this example the failed drive was already physically replaced, which is why this example uses option '-r'. Please see 'setup volume VOLNAME replace-lun -h' for details.

If this is part of a cluster, i would recommend running lunsync -r on both nodes also

---