Re: [SPDK] Why do you need cntrlid unique across all subsystems

Re: [SPDK] Why do you need cntrlid unique across all subsystems

[Walker, Benjamin]

[-- Attachment #1: Type: text/plain, Size: 985 bytes --]

On Wed, 2017-03-22 at 21:30 +0530, Sriram Popuri wrote:
> Hi experts,
> 
>    I was looking at the function spdk_nvmf_session_gen_cntlid().
> Looks like the cntlid is unique across all nvme subsystems.
> I thought the cntlid is unique in a subsystem as per the spec. Am I
> missing something here?

You're right that the specification only requires cntlid to be unique
within a subsystem. As part of an earlier effort to allocate resources
to new connections more efficiently we made the cntlid globally unique.
However, that effort didn't bear any fruit and we're now pursuing a
different, better option. The cntlid can be changed to be only unique
within a subsystem again if you or anyone wanted to create a patch.
It's not hurting anything by being globally unique, but it doesn't need
to be.

> 
> Regards,
> ~Sriram
> _______________________________________________
> SPDK mailing list
> SPDK(a)lists.01.org
> https://lists.01.org/mailman/listinfo/spdk

[-- Attachment #2: smime.p7s --]
[-- Type: application/x-pkcs7-signature, Size: 3274 bytes --]

Re: [SPDK] Why do you need cntrlid unique across all subsystems

[Sriram Popuri]

[-- Attachment #1: Type: text/plain, Size: 1548 bytes --]

Thanks for the response. I just want to understand the rationale. My
requirement would be a bit different as I need a subsystem spread across
different target systems, however I do see there are many other things in
spdk which can be contributed back.

Regards,
~Sriram

On Wed, Mar 22, 2017 at 9:52 PM, Walker, Benjamin <benjamin.walker(a)intel.com
> wrote:

> On Wed, 2017-03-22 at 21:30 +0530, Sriram Popuri wrote:
> > Hi experts,
> >
> >    I was looking at the function spdk_nvmf_session_gen_cntlid().
> > Looks like the cntlid is unique across all nvme subsystems.
> > I thought the cntlid is unique in a subsystem as per the spec. Am I
> > missing something here?
>
> You're right that the specification only requires cntlid to be unique
> within a subsystem. As part of an earlier effort to allocate resources
> to new connections more efficiently we made the cntlid globally unique.
> However, that effort didn't bear any fruit and we're now pursuing a
> different, better option. The cntlid can be changed to be only unique
> within a subsystem again if you or anyone wanted to create a patch.
> It's not hurting anything by being globally unique, but it doesn't need
> to be.
>
> >
> > Regards,
> > ~Sriram
> > _______________________________________________
> > SPDK mailing list
> > SPDK(a)lists.01.org
> > https://lists.01.org/mailman/listinfo/spdk
>
> _______________________________________________
> SPDK mailing list
> SPDK(a)lists.01.org
> https://lists.01.org/mailman/listinfo/spdk
>
>

[-- Attachment #2: attachment.html --]
[-- Type: text/html, Size: 2281 bytes --]

[SPDK] Why do you need cntrlid unique across all subsystems

[Sriram Popuri]

[-- Attachment #1: Type: text/plain, Size: 257 bytes --]

Hi experts,

   I was looking at the function spdk_nvmf_session_gen_cntlid(). Looks like
the cntlid is unique across all nvme subsystems.
I thought the cntlid is unique in a subsystem as per the spec. Am I missing
something here?

Regards,
~Sriram

[-- Attachment #2: attachment.html --]
[-- Type: text/html, Size: 469 bytes --]