[Scst-devel] [ofa-general] WinOF_2_0_5/SRP initiator: slow reads and eventually hangs
Vladislav Bolkhovitin
vst at vlnb.net
Tue Sep 15 09:39:06 PDT 2009
Chris Worley, on 09/15/2009 03:03 AM wrote:
> On Mon, Sep 14, 2009 at 12:51 PM, Vladislav Bolkhovitin <vst at vlnb.net> wrote:
>> Chris Worley, on 09/11/2009 11:50 PM wrote:
>>> I've definitely removed the switch/firmware from being the cause.
>>>
>>> I'm thinking the reason you can't repeat the test may be latency
>>> related. We get ~50usecs average latency (on small block sizes),
>>> which can't be achieved using regular SSD's (and rotating drives are
>>> nowhere close). Maybe a ramdisk would help repeat the issue.
>> I think you should try to reproduce the problem with ramdisk or nullio. By
>> so you will eliminate possible influence of the SSD backend.
>
> W/ 12GB RAM in the target, I created a 7GB ramdisk:
>
> mount -t ramfs -o size=7g ramfs /mnt/
> dd if=/dev/zero of=/mnt/foo bs=1024k count=7000
> echo "open ramdisk /mnt/foo" > /proc/scsi_tgt/vdisk/vdisk
> echo "add ramdisk 2" >/proc/scsi_tgt/groups/Default/devices
>
> Then, on the initiator, I tested it... and it hung during sequential
> 8KB block reads:
>
> fio --rw=read --bs=8k --numjobs=64 --iodepth=64 --sync=0 --direct=1
> --randrepeat=0 \
> --group_reporting --ioengine=libaio --filename=/dev/sde --name=test
> --loops=10000 --runtime=600
>
> Note that I was running the SM on the target this time too.
Should you try then with iSCSI with IPoIB? It will eliminate the SRP stack.
> Thanks,
>
> Chris
>> Vlad
>>
>
> ------------------------------------------------------------------------------
> Come build with us! The BlackBerry® Developer Conference in SF, CA
> is the only developer event you need to attend this year. Jumpstart your
> developing skills, take BlackBerry mobile applications to market and stay
> ahead of the curve. Join us from November 9-12, 2009. Register now!
> http://p.sf.net/sfu/devconf
> _______________________________________________
> Scst-devel mailing list
> Scst-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/scst-devel
>
More information about the general
mailing list