Re: How to break a mirror?
From: Mr. Johan Andersson (johan_at_solace.mh.se)
Date: 10/12/05
- Next message: Tim Bradshaw: "Re: Strange fsck behaviour with alternated superblock"
- Previous message: samedi: "Reading Long Block Record Lengths"
- In reply to: kaka.hui_at_gmail.com: "How to break a mirror?"
- Next in thread: Man Cubus: "Re: How to break a mirror?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Wed, 12 Oct 2005 10:45:48 +0200
On Wed, 11 Oct 2005, kaka.hui@gmail.com wrote:
> I have some bad disks on (d20: Submirror of d30). How do I detach that
> mirror? Thanks!
>
> [root@beta-norm01:/etc]# metastat
> d30: Mirror
> Submirror 0: d10
> State: Needs maintenance
> Submirror 1: d20
> State: Needs maintenance
> Pass: 1
> Read option: roundrobin (default)
> Write option: parallel (default)
> Size: 573315840 blocks
>
> d10: Submirror of d30
> State: Needs maintenance
> Invoke: metareplace d30 c1t34d0s6 <new device>
> Size: 573315840 blocks
> Stripe 0: (interlace: 32 blocks)
> Device Start Block Dbase State Hot Spare
> c1t32d0s6 0 No Maintenance
> c1t33d0s6 10176 No Maintenance
> c1t34d0s6 10176 No Maintenance
> c1t35d0s6 10176 No Maintenance
>
>
> d20: Submirror of d30
> State: Needs maintenance
> Invoke: after replacing "Maintenance" components:
> metareplace d30 c1t38d0s6 <new device>
> Size: 573315840 blocks
> Stripe 0: (interlace: 32 blocks)
> Device Start Block Dbase State Hot Spare
> c1t36d0s6 0 No Last Erred
> c1t37d0s6 10176 No Okay
> c1t38d0s6 10176 No Last Erred
> c1t39d0s6 10176 No Last Erred
>
>
Are you sure the disks are actually broken? This looks more like you had
some transport problems which made the systems loose contact with the
disks for enough time to put them in maintainance state.
If they are broken, replace the d10 side first, since the d20 side is
"last erred" and hence most up to date, replace one disk at a time and
let it sync up before you replace the next, use the metastat command as
above to see which to replace first, in the case above, c1t34d0s6
If they are NOT broken or if you sustpect they are not, just try to
reenable them in the order the system suggests...
i.e.
metareplace -e d30 c1t34d0s6
wait for it to sync up and watch for diskerrors.
then metastat d30
check which disk is now on the list, first d10, then when its done, d20
always in the suggested order by metastat d30
This will, if disks are really OK, go through with no faults and you
should be back to a mirrored d30, to know if you lost data or not you
would have to go back to the messagelog during the failure to see what
happened there.
My guess is that you had some kind of hickup transportwise, reason
unknown with no more info then this.
It might be evident in the logs, or it might not, since we dont even know
WHAT kind of disks it is, its difficult to say, the numbers indicate that
you might have fcal disks of some sorts, four 73Gb disks on each side it
looks like it could be :-/
/Johan A
- Next message: Tim Bradshaw: "Re: Strange fsck behaviour with alternated superblock"
- Previous message: samedi: "Reading Long Block Record Lengths"
- In reply to: kaka.hui_at_gmail.com: "How to break a mirror?"
- Next in thread: Man Cubus: "Re: How to break a mirror?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
