[ofw] Proper way to re-find mirror of SRP drives after link broken

Smith, Stan stan.smith at intel.com
Thu Jun 24 11:17:54 PDT 2010


Chris Worley wrote:
> On Thu, Jun 24, 2010 at 12:00 PM, Smith, Stan <stan.smith at intel.com>
> wrote:
>> Sufficool, Stanley wrote:
>>>> -----Original Message-----
>>>> From: ofw-bounces at lists.openfabrics.org
>>>> [mailto:ofw-bounces at lists.openfabrics.org] On Behalf Of Chris
>>>> Worley Sent: Tuesday, June 22, 2010 9:00 AM
>>>> To: ofw at lists.openfabrics.org
>>>> Subject: [ofw] ***SPAM*** Proper way to re-find mirror of SRP
>>>> drives after link broken
>>>>
>>>>
>>>> In testing SRP over windows, with two drives over SRP
>>>> mirrored, pulling then replacing the IB cable, the drives come
>>>> back as "foreign" and need to be manually imported.
>>>>
>>>> That tells me that Windows is finding something about the
>>>> return of the drives to be different than before.
>>>>
>>>> Is there a way to assure Windows sees these drives as the
>>>> drives it had before?
>>>
>>> SRP on OFED-W is broken. It sends a DREQ on driver start but AFTER
>>> the volume is mounted. So not even a basic volume will mount
>>> reliably.
>>
>>
>> Lets be clear hear w.r.t. Windows SRP.
>> I'm not disputing what your claiming w.r.t. the timing of DREQ
>> xmits, but as far as basic mounts working, they mount just fine for
>> vdisks.
>
> It does mount reliably after reboot... cable reconnect is the issue.

OK.

>
> This is a vdisk on the Linux side... I'm not sure of the vdisk you're
> referring to.

Yes - OFED-W (aka winOFED) does not support an SRP target.

>
> Are you saying there is no work-around for pulling/pushing the IB
> cable (other than manually importing then recreating the mirror)?

No at all. It's just the Windows SRP 'maintainers' are busy and have no cycles to address Windows SRP issues.
This lack of SRP support has been a major issue to a point where we are soliciting help and/or new SRP maintainers.

>
> Thanks,
>
> Chris
>>
>> stan smith.
>>
>>
>>>
>>> You will also find that this happens on a warm boot of the
>>> initiator.
>>>
>>> I'm not sure if this is your specific issue.
>>>
>>>>
>>>> Thanks,
>>>>
>>>> Chris
>>>> _______________________________________________
>>>> ofw mailing list
>>>> ofw at lists.openfabrics.org
>>>> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ofw




More information about the ofw mailing list