Bug#598793: sas controller resets causes drives to fail under mdadm
This patch has been running: while true; do smartctl -a /dev/sdb >
/dev/null; done overnight without fault, under ordinary conditions I
can reproduce within seconds.
On 02/10/10 23:56, Bjørn Mork wrote:
> which was included in Debian linux-image-2.6.32-5-amd64 version
> 2.6.32-24. You should check whether the issue is still present with
> that version.
Will do, and will let you know.
> Given the age and stability of this driver, I believe it's reasonable to
> guess that the issue was present in all 2.6.26 versions, and I guess it
> won't be backported unless you do it yourself.
I think you are correct that it has been around for all 2.6.26 versions.
I have been running this controller + kernel in the same config in the
lab for 4 months & put this controller into production 2 months ago.
During our testing period I only encountered the problem once when
completely by chance ran smartctl. I quickly learnt to avoid doing that
& it has only now reoccurred by itself in production.
I am more than happy to backport & maintain this patch out of band. It
is my impression however that the bug is not easily reproducible unless
you're running smartctl and the risk to users when it does occur can be
as severe as data loss.
Thank you for your help so far & please let me know if further testing
is required as I am happy to perform this.