zfs discuss - Aug 2008 - zpool detach from degraded mirror : why "only applicable to mirror ..." ?

Hi,

I thought that this question must have been answered already, but I have
not found any explanations. I''m sorry in advance if this is redundant,
but:

Why exactly doesn''t ZFS let me detach a device from a degraded mirror?

haggis:~# zpool  status
  pool: rmirror
 state: DEGRADED
status: One or more devices has experienced an unrecoverable error.  An
	attempt was made to correct the error.  Applications are unaffected.
action: Determine if the device needs to be replaced, and clear the errors
	using ''zpool clear'' or replace the device with
''zpool replace''.
   see: http://www.sun.com/msg/ZFS-8000-9P
 scrub: none requested
config:

	NAME         STATE     READ WRITE CKSUM
	rmirror      DEGRADED     0     0     0
          c15t0d0s0  ONLINE       0     0     0
          c16t0d0s0  DEGRADED     0     0     0  too many errors

errors: No known data errors

haggis:~# zpool detach rmirror c16t0d0s0
cannot detach c16t0d0s0: only applicable to mirror and replacing vdevs

Background: My system was crashing twice during a zpool scrub,
so I thought I should avoid a third scrubbing attemt and rather detach
and re-attach the mirror to achieve a full re-sync.

this is snv_93

Ref:

http://docs.sun.com/app/docs/doc/817-2271/gcfhe?l=en&a=view&q=%22only+applicable+to+mirror+and+replacing+vdevs%22
 
 
This message posted from opensolaris.org

2008/8/15 Nils Goroll <nils.goroll at hamburg.de>:
> Hi,
>
> I thought that this question must have been answered already, but I have
> not found any explanations. I''m sorry in advance if this is
redundant, but:
>
> Why exactly doesn''t ZFS let me detach a device from a degraded
mirror?
>
> haggis:~# zpool  status
>  pool: rmirror
>  state: DEGRADED
> status: One or more devices has experienced an unrecoverable error.  An
>        attempt was made to correct the error.  Applications are unaffected.
> action: Determine if the device needs to be replaced, and clear the errors
>        using ''zpool clear'' or replace the device with
''zpool replace''.
>   see: http://www.sun.com/msg/ZFS-8000-9P
>  scrub: none requested
> config:
>
>        NAME         STATE     READ WRITE CKSUM
>        rmirror      DEGRADED     0     0     0
>          c15t0d0s0  ONLINE       0     0     0
>          c16t0d0s0  DEGRADED     0     0     0  too many errors
>
That is not a mirror. This is an example of a mirror (3-way):

        NAME         STATE     READ WRITE CKSUM
        rpool         ONLINE       0     0     0
          mirror      ONLINE       0     0     0
            c1d0s0    ONLINE       0     0     1
            c2d0s0    ONLINE       0     0     1
            c4t0d0s0  ONLINE       0     0     0

The fact that it''s DEGRADED and not FAULTED indicates that it thinks
the
DTL (dirty time logs) for the two sides of the mirrors overlap in some
way, so detaching it would result in loss of data.  In the process of
doing this, it seems the error message got lost, and you ended up with
something completely non-sensical.  If you do a ''zpool scrub'',
does it
complete without any errors?

- Eric

On Fri, Aug 15, 2008 at 01:48:48PM -0700, Nils Goroll
wrote:
> Hi,
> 
> I thought that this question must have been answered already, but I have
> not found any explanations. I''m sorry in advance if this is
redundant, but:
> 
> Why exactly doesn''t ZFS let me detach a device from a degraded
mirror?
> 
> haggis:~# zpool  status
>   pool: rmirror
>  state: DEGRADED
> status: One or more devices has experienced an unrecoverable error.  An
> 	attempt was made to correct the error.  Applications are unaffected.
> action: Determine if the device needs to be replaced, and clear the errors
> 	using ''zpool clear'' or replace the device with
''zpool replace''.
>    see: http://www.sun.com/msg/ZFS-8000-9P
>  scrub: none requested
> config:
> 
> 	NAME         STATE     READ WRITE CKSUM
> 	rmirror      DEGRADED     0     0     0
>           c15t0d0s0  ONLINE       0     0     0
>           c16t0d0s0  DEGRADED     0     0     0  too many errors
> 
> errors: No known data errors
> 
> haggis:~# zpool detach rmirror c16t0d0s0
> cannot detach c16t0d0s0: only applicable to mirror and replacing vdevs
> 
> Background: My system was crashing twice during a zpool scrub,
> so I thought I should avoid a third scrubbing attemt and rather detach
> and re-attach the mirror to achieve a full re-sync.
> 
> this is snv_93
> 
> Ref:
> 
>
http://docs.sun.com/app/docs/doc/817-2271/gcfhe?l=en&a=view&q=%22only+applicable+to+mirror+and+replacing+vdevs%22
>  
>  
> This message posted from opensolaris.org
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
--
Eric Schrock, Fishworks                        http://blogs.sun.com/eschrock

On Fri, Aug 15, 2008 at 02:14:02PM -0700, Eric Schrock
wrote:
> The fact that it''s DEGRADED and not FAULTED indicates that it
thinks the
> DTL (dirty time logs) for the two sides of the mirrors overlap in some
> way, so detaching it would result in loss of data.  In the process of
> doing this, it seems the error message got lost, and you ended up with
> something completely non-sensical.  If you do a ''zpool
scrub'', does it
> complete without any errors?
Ooops, points to Mattias to paying more attention than I was :-)

- Eric

--
Eric Schrock, Fishworks                        http://blogs.sun.com/eschrock

Matthias,

that does not answer my question.

The question is: Why can''t I decide that I consciously want to destroy
the (two way)
mirror (and, yes, do away with any redundancy).

Nils
 
 
This message posted from opensolaris.org

On Saturday, August 16, 2008 at 00:05:17 CEST, Nils Goroll
wrote:
> Matthias,
> 
> that does not answer my question.
> 
> The question is: Why can''t I decide that I consciously want to
destroy the (two way)
> mirror (and, yes, do away with any redundancy).
> 
Hi,

this pool does not have any redundancy because it is not a mirror.
Hence you can''t detach anything from it. The pool is named
"rmirror".
That does not make it a mirror.
>       NAME         STATE     READ WRITE CKSUM
>       rmirror      DEGRADED     0     0     0
>           c15t0d0s0  ONLINE       0     0     0
>           c16t0d0s0  DEGRADED     0     0     0  too many errors
This is an unmirrored pool consisting of two vdevs: c15t0d0s0 and c16t0d0s0.

Ronald

Hi all, especially Matthias,

I am very sorry for having bothered you with this stupid question, I am
embarrassed by the fact that I did not realize it''s not a mirror. The
fact that I named it "rmirror" definitely added confusion on my side.

Apologies in particular for not having taken Mathias'' initial hint for
what it was.

By the way, I could finally scrub the pool, no problems remaining
(except for the task to make it a real mirror).

Thanks again,

Nils
 
 
This message posted from opensolaris.org