The changes are in multipath_error(). The diffs for /derivers/md/multipath.c
are:
233,234c233
< /* leave it active... it's all we have */
< } else {
---
}
252c251
< }
---
Thanks.
----- Original Message -----
From: "NeilBrown" <neilb@xxxxxxx>
To: "Ron Lai" <ronlai@xxxxxxxxxxxxxxx>
Cc: <linux-raid@xxxxxxxxxxxxxxx>
Sent: Thursday, June 25, 2009 6:27 PM
Subject: Re: Questions regarding failure in the last IO path in multipath
On Fri, June 26, 2009 10:30 am, Ron Lai wrote:
Hi,
I use mdadm to setup multipath on a box running 2.6.25.14 and I try to
simualte SAN failure by shutting down ports on the switch one by one.
After
I disable the last IO path, the box hangs with the following logs spewed
out
on the console.
Jun 23 14:32:31 localhost kernel: multipath: sdn2: rescheduling sector 8
Jun 23 14:32:31 localhost kernel: multipath: sdn: redirecting sector 48 to
another IO path
Jun 23 14:32:31 localhost kernel: multipath: sdn: redirecting sector
200015880 to another IO path
Jun 23 14:32:31 localhost kernel: multipath: sdn: redirecting sector
200016408 to another IO path
Jun 23 14:32:31 localhost kernel: multipath: sdn: redirecting sector
200114976 to another IO path
Jun 23 14:32:31 localhost kernel: multipath: only one IO path left and IO
error.
Jun 23 14:32:31 localhost kernel: multipath: sdn2: rescheduling sector 56
Jun 23 14:32:31 localhost kernel: multipath: only one IO path left and IO
error.
Jun 23 14:32:31 localhost kernel: multipath: sdn2: rescheduling sector
200015888
Jun 23 14:32:31 localhost kernel: multipath: only one IO path left and IO
error.
Jun 23 14:32:31 localhost kernel: multipath: sdn2: rescheduling sector
200016416
Jun 23 14:32:31 localhost kernel: multipath: only one IO path left and IO
error.
Jun 23 14:32:31 localhost kernel: multipath: sdn2: rescheduling sector
200114984
Jun 23 14:32:31 localhost kernel: multipath: sdn: redirecting sector 0 to
another IO path
Jun 23 14:32:31 localhost kernel: multipath: sdn: redirecting sector 48 to
another IO path
Jun 23 14:32:31 localhost kernel: multipath: sdn: redirecting sector
200015880 to another IO path
Jun 23 14:32:31 localhost kernel: multipath: sdn: redirecting sector
200016408 to another IO path
Jun 23 14:32:31 localhost kernel: multipath: only one IO path left and IO
error.
Jun 23 14:32:31 localhost kernel: multipath: sdn2: rescheduling sector 8
Jun 23 14:32:31 localhost kernel: multipath: only one IO path left and IO
error.
Jun 23 14:32:31 localhost kernel: multipath: sdn2: rescheduling sector 56
Jun 23 14:32:31 localhost kernel: multipath: only one IO path left and IO
error.
Jun 23 14:32:31 localhost kernel: multipath: sdn2: rescheduling sector
200015888
Jun 23 14:32:31 localhost kernel: multipath: only one IO path left and IO
error.
Jun 23 14:32:32 localhost kernel: multipath: sdn2: rescheduling sector
200016416
Jun 23 14:32:32 localhost kernel: multipath: sdn: redirecting sector
200114976 to another IO path
........
In multipath_error() in /derivers/md/multipath.c the code says that the
last
path never gets marked as faulty hence it keeps retrying. I've tried to
change the code to allow the last path to be marked as faulty and the box
seems to work okay after the failure in the last path. (The kernel spews
out
logs about inode errors. Nevertheless it stops eventually and mdadm shows
that all the paths are marked as faulty.) Will this modification create
problems in other parts of the kernel? Thanks.
Probably not, but it is hard to tell unless you post the actual patch.
Could you do that please?
Thanks,
NeilBrown
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
