Is it possible to downgrade a 3-disk mirror to 2-disk or the revers?

[Jimm Chen]

I find FreeNAS uniqe to allow 3-disk mirror to make a very robust redundancy and 3X read speed. But once I've made the decision to create a 3-disk mirror on creating a zpool, can I later downgrade it to 2-disk mirror? Simply pull one member disk offline(via WebUI) does not do the trick because FreeNAS will report my pool is in degraded state, forever.

I'm trying FreeNAS-9.3-STABLE-201504100216 , but seems not able to find the way via FreeNAS WebUI. Then does some FreeBSD command line exist to do that?

BTW: What does the 5577469217091291914 serial mean for a offline disk? Where else can I see this serial?

Thank you.

 

[Ericloewe]

You can do this, but only from the CLI. Simply pulling a drive does not work and is not supposed to work.

 

[Robert Trevellyan]

Then does some FreeBSD command line exist to do that?

The command you need to read about is zpool detach.

 

[Jimm Chen]

Thank you, Robert. You're right. My output below:

Code:

[ /dev/pts/1(screen-256color) 2015-05-11 17:20:22 jobs:0 ERR:0]
[ root @freenas93 /dev/zvol/zpool1 ]
# zpool status zpool1
  pool: zpool1
state: ONLINE
  scan: resilvered 4.43M in 0h0m with 0 errors on Mon Apr 27 16:03:13 2015
config:

        NAME                                            STATE     READ WRITE CKSUM
        zpool1                                          ONLINE       0     0     0
          mirror-0                                      ONLINE       0     0     0
            gptid/9a9e85e5-e348-11e4-9557-000c2996193e  ONLINE       0     0     0
            gptid/9b09a590-e348-11e4-9557-000c2996193e  ONLINE       0     0     0
            gptid/9b757f00-e348-11e4-9557-000c2996193e  ONLINE       0     0     0

errors: No known data errors

[ /dev/pts/1(screen-256color) 2015-05-11 17:26:44 jobs:0 ERR:0]
[ root @freenas93 /dev/zvol/zpool1 ]
# zpool detach zpool1 gptid/9a9e85e5-e348-11e4-9557-000c2996193e

[ /dev/pts/1(screen-256color) 2015-05-11 17:27:45 jobs:0 ERR:0]
[ root @freenas93 /dev/zvol/zpool1 ]
# zpool status zpool1
  pool: zpool1
state: ONLINE
  scan: resilvered 4.43M in 0h0m with 0 errors on Mon Apr 27 16:03:13 2015
config:

        NAME                                            STATE     READ WRITE CKSUM
        zpool1                                          ONLINE       0     0     0
          mirror-0                                      ONLINE       0     0     0
            gptid/9b09a590-e348-11e4-9557-000c2996193e  ONLINE       0     0     0
            gptid/9b757f00-e348-11e4-9557-000c2996193e  ONLINE       0     0     0

errors: No known data errors

But there is one pending question, how do I know the mapping relationship between a lengthy identifier(like gptid/9a9e85e5-e348-11e4-9557-000c2996193e) and hard disk device name(da0, da1 etc) .

 

[Robert Trevellyan]

I believe @Bidule0hm has published some useful scripts that at least can map gptid to serial #. Maybe they show gptid too. If not, the View Disks button shows your adax -> serial # mapping.

 

[Bidule0hm]

Yep, it maps the GPTID with the serial and the device name (but I recommend to never use the device name to identify a drive from reboot to reboot because it can change, it's only here as it can be useful for edge cases and if you know what you're doing), see the useful scripts link in my signature ;)

 

[Robert Trevellyan]

how do I know the mapping relationship between a lengthy identifier(like gptid/9a9e85e5-e348-11e4-9557-000c2996193e) and hard disk device name(da0, da1 etc) .

One more thing ... don't fall into the trap of assuming that the device numbers (da0 etc) correspond to the port numbers on your motherboard or HBA. Many times they will, but it is far from guaranteed.