9.10 non-native block size error

[rosabox]

I have strange problem
I have a freshly installed FreeNAS 9.10 (upgraded yesterday to 9.10.1), I created a raidz2 pool from 8 disks, all of them with 512b sectors.
I checked ashift (it's 12), because I'm planning to replace the disks with bigger ones which will have 4k sectors.
Today I started replacing the first one and when I check zpool status, I get this:

Code:

  pool: tank
state: ONLINE
status: One or more devices is currently being resilvered.  The pool will
        continue to function, possibly in a degraded state.
action: Wait for the resilver to complete.
  scan: resilver in progress since Fri Aug  5 16:14:29 2016
        1.60T scanned out of 6.89T at 260M/s, 5h55m to go
        205G resilvered, 23.26% done
config:

        NAME                                              STATE     READ WRITE CKSUM
        tank                                              ONLINE       0     0     0
          raidz2-0                                        ONLINE       0     0     0
            gptid/6b0baa9f-4468-11e6-a9c8-d05099c0ea27    ONLINE       0     0     0
            gptid/6be90cce-4468-11e6-a9c8-d05099c0ea27    ONLINE       0     0     0
            gptid/6cc469c2-4468-11e6-a9c8-d05099c0ea27    ONLINE       0     0     0
            gptid/6e65e3af-4468-11e6-a9c8-d05099c0ea27    ONLINE       0     0     0
            gptid/7013e7b6-4468-11e6-a9c8-d05099c0ea27    ONLINE       0     0     0
            gptid/71151d5b-4468-11e6-a9c8-d05099c0ea27    ONLINE       0     0     0
            gptid/72b658b5-4468-11e6-a9c8-d05099c0ea27    ONLINE       0     0     0
            replacing-7                                   ONLINE       0     0     0
              gptid/746696c9-4468-11e6-a9c8-d05099c0ea27  ONLINE       0     0     0
              gptid/eb8b8019-5b16-11e6-a0b2-d05099c0ea27  ONLINE       0     0     0  block size: 512B configured, 4096B native  (resilvering)

errors: No known data errors

I thought that, when I have ashift=12, I'm ready for 4k sector disks.
Did I miss something?

 

[rosabox]

Some additional info, camcontrol identify adaX | grep sector for the old disks:
sector size logical 512, physical 512, offset 0
and for the new replacement disk:
sector size logical 512, physical 4096, offset 0

Is it a bug? If yes, then I'll fill bug report
I had the impression, from reading the docs and the forum, that FreeNAS 9.10 assumes 4k sector disk and sets ashift to 12, which is (should be) the right setting for 4k disks. Which leads me to the conclusion I've set it correct and the warning is incorrect. But I would like to rule out, that I misunderstood something.

I would really appreciate some advice.

 

[Bidule0hm]

Looks like your pool doesn't have ashift = 12 but ashift = 9. Can you post the output of zdb -C your_pool_name | grep ashift please?

 

[rosabox]

First, thank you for your answer.

I get this:

Code:

zdb -C tank | grep ashift
zdb: can't open 'tank': No such file or directory

and

Code:

zdb -C freenas-boot | grep ashift
  ashift: 12

but

Code:

zpool list
NAME           SIZE  ALLOC   FREE  EXPANDSZ   FRAG    CAP  DEDUP  HEALTH  ALTROOT
freenas-boot  59.5G  1.61G  57.9G         -      -     2%  1.00x  ONLINE  -
tank          10.9T  6.92T  3.96T         -    34%    63%  1.00x  ONLINE  /mnt

Code:

zpool status freenas-boot
  pool: freenas-boot
state: ONLINE
  scan: scrub repaired 0 in 0h0m with 0 errors on Fri Aug  5 03:45:09 2016
config:

        NAME                                          STATE     READ WRITE CKSUM
        freenas-boot                                  ONLINE       0     0     0
          gptid/64645315-35a4-11e6-8e80-d05099c0ea27  ONLINE       0     0     0

errors: No known data errors

Code:

zpool status tank
  pool: tank
state: ONLINE
status: One or more devices are configured to use a non-native block size.
  Expect reduced performance.
action: Replace affected devices with devices that support the
  configured block size, or migrate data to a properly configured
  pool.
  scan: resilvered 882G in 7h50m with 0 errors on Sat Aug  6 00:04:57 2016
config:

  NAME  STATE  READ WRITE CKSUM
  tank  ONLINE  0  0  0
  raidz2-0  ONLINE  0  0  0
  gptid/6b0baa9f-4468-11e6-a9c8-d05099c0ea27  ONLINE  0  0  0
  gptid/6be90cce-4468-11e6-a9c8-d05099c0ea27  ONLINE  0  0  0
  gptid/6cc469c2-4468-11e6-a9c8-d05099c0ea27  ONLINE  0  0  0
  gptid/6e65e3af-4468-11e6-a9c8-d05099c0ea27  ONLINE  0  0  0
  gptid/7013e7b6-4468-11e6-a9c8-d05099c0ea27  ONLINE  0  0  0
  gptid/71151d5b-4468-11e6-a9c8-d05099c0ea27  ONLINE  0  0  0
  gptid/72b658b5-4468-11e6-a9c8-d05099c0ea27  ONLINE  0  0  0
  gptid/eb8b8019-5b16-11e6-a0b2-d05099c0ea27  ONLINE  0  0  0  block size: 512B configured, 4096B native

errors: No known data errors

It seems something is wrong with my pool but what?

 

[Bidule0hm]

Ok, now I'm puzzled. Anyone?

 

[rosabox]

Meanwhile I was googling a little bit and found this command:

Code:

zdb -l /dev/ada1p2 | grep ashift
        ashift: 9
        ashift: 9
        ashift: 9
        ashift: 9

So my pool has indeed ashift=9 but why and how to fix it?

 

[Bidule0hm]

When did you create this pool?

AFAIK you can't change the ashift once the pool is created, you need to destroy and recreate it.

 

[rosabox]

I created the pool 2-3 months ago under FreeNAS 9.10

I know I have to destroy and recreate the pool.
The question is - how to make it right this time?
Because I was pretty sure I did everything correct the last time...

 

[Bidule0hm]

Maybe it's a bug in FreeNAS 9.10. On old versions the ashift was 9, it then was changed to 12 and I know it is still at 12 on FreeNAS 9.3. Maybe it has changed back to 9 on 9.10 for whatever reason. I don't know how you can set the ashift, I looked in the manual and there's nothing about it (and I remember that before we were able to change it manually, now the option seems to have disappeared).

@cyberjock @jgreco any ideas?

 

[cyberjock]

I'm pretty sure this is a bug in 9.10, but I haven't looked into this issue heavily.

 

[rosabox]

It looks like a bug to me, too.
I'm going to fill a bug report.

Edit: Done - https://bugs.freenas.org/issues/16819

 

[Chris Moore]

I just edited the first post. The disks are assumed to be 4k disks by FreeNAS starting with 9.2.0-RELEASE. So if your ashift=9 you will get the error even if its is "correct". Correct being ashift=9 and your disks are 512bytes/sector disks.

Woohoo for more confusion!

5 pages of posts, no I didn't read them all. If it is supposed to default to ashift=12, why is a new pool created in version 9.10 giving me an ashift=9?

 

[rosabox]

That I would like to know too and how to force ashift=12?

 

[rosabox]

The thread continues here: ...
It was split in some strange way.

 

[wblock]

Two pieces of thread merged. Sorry, my fault.