* raid repair checkpoint enhancement
@ 2012-10-04 17:26 Ryan, MichaelS (ENG Redmond)
2012-10-04 21:05 ` NeilBrown
0 siblings, 1 reply; 2+ messages in thread
From: Ryan, MichaelS (ENG Redmond) @ 2012-10-04 17:26 UTC (permalink / raw)
To: linux-raid@vger.kernel.org
Hi all.
If a device reboots during a raid repair operation ("echo repair > /sys/block/md0/md/sync_action"), the raid repair is interrupted and not restarted. For devices that are shutdown nightly, this could prevent a raid repair from ever getting to the far reaches of the array. I'd like to see an enhancement that would checkpoint the raid repair operation on shutdown similar to how a sync operation is checkpointed. Preferably, the raid repair would be restarted automatically on the next boot at the checkpoint location. Alternatively it could require user input to restart the operation at either the checkpoint location, or at the beginning.
Any thoughts on this enhancement?
Thanks
- Michael
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: raid repair checkpoint enhancement
2012-10-04 17:26 raid repair checkpoint enhancement Ryan, MichaelS (ENG Redmond)
@ 2012-10-04 21:05 ` NeilBrown
0 siblings, 0 replies; 2+ messages in thread
From: NeilBrown @ 2012-10-04 21:05 UTC (permalink / raw)
To: Ryan, MichaelS (ENG Redmond); +Cc: linux-raid@vger.kernel.org
[-- Attachment #1: Type: text/plain, Size: 1074 bytes --]
On Thu, 4 Oct 2012 13:26:35 -0400 "Ryan, MichaelS (ENG Redmond)"
<michaels.ryan@emc.com> wrote:
>
> Hi all.
>
> If a device reboots during a raid repair operation ("echo repair > /sys/block/md0/md/sync_action"), the raid repair is interrupted and not restarted. For devices that are shutdown nightly, this could prevent a raid repair from ever getting to the far reaches of the array. I'd like to see an enhancement that would checkpoint the raid repair operation on shutdown similar to how a sync operation is checkpointed. Preferably, the raid repair would be restarted automatically on the next boot at the checkpoint location. Alternatively it could require user input to restart the operation at either the checkpoint location, or at the beginning.
>
> Any thoughts on this enhancement?
You can do this yourself with some scripting.
Before you shut down, record the first number in
/sys/block/mdXX/md/sync_completed
When on reboot, write that number to
/sys/block/mdXX/md/sync_min
before writing 'repair' to 'sync_action'.
NeilBrown
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 828 bytes --]
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2012-10-04 21:05 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-10-04 17:26 raid repair checkpoint enhancement Ryan, MichaelS (ENG Redmond)
2012-10-04 21:05 ` NeilBrown
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).