[DRBD-user] Re: Poor network performance on 0.7.22

Oliver Hookins oliver.hookins at anchor.com.au
Mon Jun 23 09:43:33 CEST 2008


On Mon Jun 23, 2008 at 09:27:02 +0200, Lars Ellenberg wrote:
>On Mon, Jun 23, 2008 at 02:50:07PM +1000, Oliver Hookins wrote:
>> Yet more information on this problem. I've found if I remove one of the
>> disks from the software RAID1 on the Secondary in this DRBD resource, the
>> transfer speeds jump up to what I would call acceptable levels.
>> 
>> I've tested each of the disks independently and they are both fine. It seems
>> there is some sort of strange interaction going on between DRBD, the md
>> layer, and perhaps the Adaptec 2020ZCR card (but I'm not sure). With so many
>> layers of I/O abstraction it is hard to know what is going on and what
>> particular thing is causing the problem...
>> 
>> Any bells ringing for anyone?
>
>would this perhaps help a little?
>git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
>
>commit e3881a6816b45668df60a426e5c3431ece1539a7
>Author: Lars Ellenberg <Lars.Ellenberg at linbit.com>
>Date:   Wed Jan 10 23:15:37 2007 -0800
>
>    [PATCH] md: pass down BIO_RW_SYNC in raid{1,10}
>    
>    md raidX make_request functions strip off the BIO_RW_SYNC flag, thus
>    introducing additional latency.
>    
>    Fixing this in raid1 and raid10 seems to be straightforward enough.
>    
>    For our particular usage case in DRBD, passing this flag improved some
>    initialization time from ~5 minutes to ~5 seconds.
>    
>    Acked-by: NeilBrown <neilb at suse.de>
>    Signed-off-by: Lars Ellenberg <lars at linbit.com>
>    Acked-by: Jens Axboe <jens.axboe at oracle.com>
>    Cc: <stable at kernel.org>
>    Signed-off-by: Andrew Morton <akpm at osdl.org>
>    Signed-off-by: Linus Torvalds <torvalds at osdl.org>

Indeed it would. I'll have to check and see if it was backported to the
RHEL4 kernel (unless you know off the top of your head).

-- 
Regards,
Oliver Hookins



More information about the drbd-user mailing list