Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Thu, Aug 13, 2009 at 06:07:30PM +0200, hans morbus2000 wrote:
> hello,
>
> we add the following parameter in drbd.conf
>
> disk {
> ...
> max-bio-bvecs 1;
fine.
> ...
> }
>
> and ...
>
> # echo 4 > /sys/block/sdb/queue/max_sectors_kb
which is not neccessary.
> now, the performance of the 3ware 9550 is very slow
>
> # hdparm -tT /dev/sda
> /dev/sda:
> Timing cached reads: 8472 MB in 2.00 seconds = 4246.06 MB/sec
> Timing buffered disk reads: 130 MB in 3.04 seconds = 42.81 MB/sec
>
> I've applied the standard 3ware tuning values
> echo 128 > /sys/block/sda/queue/max_sectors_kb (default 128)
> /sbin/blockdev --setra 16384 /dev/sda
>
> and the performance goes to
>
> # hdparm -tT /dev/sda
> /dev/sda:
> Timing cached reads: 8152 MB in 2.00 seconds = 4083.90 MB/sec
> Timing buffered disk reads: 828 MB in 3.01 seconds = 275.36 MB/sec
>
> can we change backwards max_sectors_kb to 128
> without to get the same error?
Yes.
but that "performance" does not really come from max_sectors_kb,
but mainly from the readahead.
which I really recommend against, unless you do streaming media ;)
because that is good for streaming read micro benchmarks,
but seriously hurts perfromance for random IO, and stresses the VM
and pagecache unneccessarily.
but hey, just do as you please.
--
: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com
DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
__
please don't Cc me, but send to list -- I'm subscribed