|
Unfortunately, I can't make that slot today - I've got an internal status call today that I have to run...
Ryan
controller-dev-bounces@... wrote on 01/13/2015 12:13:01 PM:
> From: Kyle Mestery <mestery@...>
> To: Colin Dixon <colin@...>, Ryan Moats <regxboi@...>
> Cc: "<discuss@...>"
> <discuss@...>, plugin2oc-
> dev@..., "<ovsdb-dev@...>"
> <ovsdb-dev@...>, opendove-
> dev@..., controller-dev <controller-
> dev@...>, Ed Warnicke <eaw@...>, "vtn-
> dev@..." <vtn-dev@...>
> Date: 01/13/2015 12:13 PM
> Subject: Re: [controller-dev] [OpenDaylight Discuss] MD-SAL and the
> future of Neutron
> Sent by: controller-dev-bounces@...
>
> My opinion is lets use the OVSDB call for this. Do we plan to do
> that today? I want to ensure that Ryan and I are at this meeting
> when it happens. As we've discussed in [1], I'm in favor of anything
> which does not change the interface to Neutron on the NB side.
>
> On Tue, Jan 13, 2015 at 11:51 AM, Colin Dixon <colin@...> wrote:
> Thanks! I'm also cc'ing the right list for plugin2oc here.
>
> --Colin
>
> On Tue, Jan 13, 2015 at 8:20 AM, Reinaldo Penno <rapenno@...> wrote:
> Let me know about the call or any work items and I will pitch in. I
> have been looking into Neutron integration can really use a revamped
> Neutron API in SFC.
>
> From: Colin Dixon <colin@...>
> Date: Tuesday, January 13, 2015 at 8:11 AM
> To: "<ovsdb-dev@...>" <ovsdb-dev@...
> >, controller-dev <controller-dev@...>, <
> plugin2oc-dev@...>, <opendove-dev@...>, "
> vtn-dev@..." <vtn-dev@...>, "<
> discuss@...>" <discuss@...>
> Cc: Ed Warnicke <eaw@...>, Anton Ivanov <anton.ivanov@...>
> Subject: [controller-dev] MD-SAL and the future of Neutron
>
> Recently, there's been some movement [0] (mainly from Greg who is
> cc'd) on trying to provide MD-SAL models for the Neutron API and how
> to eventually migrate toward having our Neutron APIs be more of a
> first class MD-SAL citizen. After that Ed and Ryan have expressed
> some opinions [1] about how we should approach this.
>
> Given that we have at least 4 projects that implement the Neutron
> interface and it is one of the most important features ODL offers—if
> not *the* most important feature, it would behoove us to get this right.
> Given that, I'd really like to see the stakeholders come together
> and try to work out how we can move in the right direction. I've
> sent this to the 4 project -dev lists, the controller-dev list
> (where the API is housed at the moment) and to discuss, so I don't
> think I've missed anyone.
> If people think it's useful to set up a series of calls for this, or
> just use the OVSDB call [2] (which I think is the only project with
> a regularly scheduled meeting—I think that would be useful.
>
> --Colin
>
> [0] https://lists.opendaylight.org/pipermail/discuss/2014-December/004141.html
> [1] https://lists.opendaylight.org/pipermail/discuss/2015-January/004207.html
> [2] https://wiki.opendaylight.org/view/
> OVSDB_Integration:Main#Weekly_OVSDB_Plugin_Meeting_.28Open_to_All.29
> _______________________________________________ controller-dev mailing list
> controller-dev@... https://
> lists.opendaylight.org/mailman/listinfo/controller-dev
>
>
> _______________________________________________
> Discuss mailing list
> Discuss@...
> https://lists.opendaylight.org/mailman/listinfo/discuss
> _______________________________________________
> controller-dev mailing list
> controller-dev@...
> https://lists.opendaylight.org/mailman/listinfo/controller-dev
|
|
|
Kyle Mestery <mestery@...>
Should we postpone this to next week?
On Tue, Jan 13, 2015 at 12:19 PM, Ryan Moats <rmoats@...> wrote:
Unfortunately, I can't make that slot today - I've got an internal status call today that I have to run...
Ryan
controller-dev-bounces@... wrote on 01/13/2015 12:13:01 PM:
> From: Kyle Mestery <mestery@...>
> To: Colin Dixon <colin@...>, Ryan Moats <regxboi@...>
> Cc: "<discuss@...>"
> <discuss@...>, plugin2oc-
> dev@..., "<ovsdb-dev@...>"
> <ovsdb-dev@...>, opendove-
> dev@..., controller-dev <controller-
> dev@...>, Ed Warnicke <eaw@...>, "vtn-
> dev@..." <vtn-dev@...>
> Date: 01/13/2015 12:13 PM
> Subject: Re: [controller-dev] [OpenDaylight Discuss] MD-SAL and the
> future of Neutron
> Sent by: controller-dev-bounces@...
>
> My opinion is lets use the OVSDB call for this. Do we plan to do
> that today? I want to ensure that Ryan and I are at this meeting
> when it happens. As we've discussed in [1], I'm in favor of anything
> which does not change the interface to Neutron on the NB side.
>
> On Tue, Jan 13, 2015 at 11:51 AM, Colin Dixon <colin@...> wrote:
> Thanks! I'm also cc'ing the right list for plugin2oc here.
>
> --Colin
>
> On Tue, Jan 13, 2015 at 8:20 AM, Reinaldo Penno <rapenno@...> wrote:
> Let me know about the call or any work items and I will pitch in. I
> have been looking into Neutron integration can really use a revamped
> Neutron API in SFC.
>
> From: Colin Dixon <colin@...>
> Date: Tuesday, January 13, 2015 at 8:11 AM
> To: "<ovsdb-dev@...>" <ovsdb-dev@...
> >, controller-dev <controller-dev@...>, <
> plugin2oc-dev@...>, <opendove-dev@...>, "
> vtn-dev@..." <vtn-dev@...>, "<
> discuss@...>" <discuss@...>
> Cc: Ed Warnicke <eaw@...>, Anton Ivanov <anton.ivanov@...>
> Subject: [controller-dev] MD-SAL and the future of Neutron
>
> Recently, there's been some movement [0] (mainly from Greg who is
> cc'd) on trying to provide MD-SAL models for the Neutron API and how
> to eventually migrate toward having our Neutron APIs be more of a
> first class MD-SAL citizen. After that Ed and Ryan have expressed
> some opinions [1] about how we should approach this.
>
> Given that we have at least 4 projects that implement the Neutron
> interface and it is one of the most important features ODL offers—if
> not *the* most important feature, it would behoove us to get this right.
> Given that, I'd really like to see the stakeholders come together
> and try to work out how we can move in the right direction. I've
> sent this to the 4 project -dev lists, the controller-dev list
> (where the API is housed at the moment) and to discuss, so I don't
> think I've missed anyone.
> If people think it's useful to set up a series of calls for this, or
> just use the OVSDB call [2] (which I think is the only project with
> a regularly scheduled meeting—I think that would be useful.
>
> --Colin
>
> [0] https://lists.opendaylight.org/pipermail/discuss/2014-December/004141.html
> [1] https://lists.opendaylight.org/pipermail/discuss/2015-January/004207.html
> [2] https://wiki.opendaylight.org/view/
> OVSDB_Integration:Main#Weekly_OVSDB_Plugin_Meeting_.28Open_to_All.29
> _______________________________________________ controller-dev mailing list
> controller-dev@... https://
> lists.opendaylight.org/mailman/listinfo/controller-dev
>
>
> _______________________________________________
> Discuss mailing list
> Discuss@...
> https://lists.opendaylight.org/mailman/listinfo/discuss
> _______________________________________________
> controller-dev mailing list
> controller-dev@...
> https://lists.opendaylight.org/mailman/listinfo/controller-dev
|
|
|
Edward Warnicke <hagbard@...>
As this issue transcends many projects across many timezones, and noon PST is almost the worst possible time for many such folks, could we get a dedicated ODL neutron callat a reasonable time, like Tuesday 8am PST that could be reasonable for all parties involved?
Ed
|
|
|
Kyle Mestery <mestery@...>
That time won't work, it overlaps with the upstream Neutron meeting. Given that the OVSDB project is driving all the Neutron work, and we've been working the plan to address things there, I think it makes sense to use that meeting. If the timing of that meeting won't work for everyone in other timezones, I propose we alternate that meeting.
On Tue, Jan 13, 2015 at 12:50 PM, Edward Warnicke <hagbard@...> wrote: As this issue transcends many projects across many timezones, and noon PST is almost the worst possible time for many such folks, could we get a dedicated ODL neutron callat a reasonable time, like Tuesday 8am PST that could be reasonable for all parties involved?
Ed
|
|
|
Edward Warnicke <hagbard@...>
I think we need to be sensitive to the fact that we *are* a community, and not just a single project, and that there are broad community interests in this and they should be addressed by the community, not just a single project.
Ed
On Tue, Jan 13, 2015 at 12:53 PM, Kyle Mestery <mestery@...> wrote: That time won't work, it overlaps with the upstream Neutron meeting. Given that the OVSDB project is driving all the Neutron work, and we've been working the plan to address things there, I think it makes sense to use that meeting. If the timing of that meeting won't work for everyone in other timezones, I propose we alternate that meeting.
|
|
|
Kyle Mestery <mestery@...>
Great sentiment! In a similar manner, I think we should also be sensitive to the people who have been working on this for two years, and are currently driving to fix things from both sides. I'm happy to open this up more, but at the same time, lets not act like no one has been looking at this. People are actively hammering on this, and in fact we've just recently been happy to welcome a plethora of new people to this team.
But I'm curious here: What parts need to bubble up past OVSDB? Because the parts above it are just passing through the Neutron APIs. The actual implementation of how things are programmed on the hosts running the VMs is all done by OVSDB.
On Tue, Jan 13, 2015 at 1:09 PM, Edward Warnicke <hagbard@...> wrote: I think we need to be sensitive to the fact that we *are* a community, and not just a single project, and that there are broad community interests in this and they should be addressed by the community, not just a single project.
Ed
|
|
|
OVSDB project was the first and it is still the reference for OpenStack-OpenDaylight integration so I personally do not see any issue in using their call to talk about Neutron strategy. Moreover I was going to ask the OVSDB group whether they could help with some of the Neutron coordination tasks I think we need for this release, like:
- Maintaining a fluent conversation with OpenStack community to better understand OS requirements for ODL as well as to keep OS community updated with ODL progress
- Doing some coordination among the OpenDaylight projects consuming the Neutron interface so maybe one day we do not need to shutdown all except one
- Updating and maintaining the OpenDaylight Neutron interface and the corresponding driver in OpenStack
On Jan 13, 2015, at 11:16 AM, Kyle Mestery < mestery@...> wrote:
Great sentiment! In a similar manner, I think we should also be sensitive to the people who have been working on this for two years, and are currently driving to fix things from both sides. I'm happy to open this up more, but at the same time, lets not act like no one has been looking at this. People are actively hammering on this, and in fact we've just recently been happy to welcome a plethora of new people to this team.
But I'm curious here: What parts need to bubble up past OVSDB? Because the parts above it are just passing through the Neutron APIs. The actual implementation of how things are programmed on the hosts running the VMs is all done by OVSDB.
|
|
|
Sorry just a precision: I think the first OpenDaylight project implementing Neutron interface was OpenDove, but unfortunately they are not with us anymore.
BR/Luis
On Jan 13, 2015, at 9:04 PM, Luis Gomez < ecelgp@...> wrote:
OVSDB project was the first and it is still the reference for OpenStack-OpenDaylight integration so I personally do not see any issue in using their call to talk about Neutron strategy. Moreover I was going to ask the OVSDB group whether they could help with some of the Neutron coordination tasks I think we need for this release, like:
- Maintaining a fluent conversation with OpenStack community to better understand OS requirements for ODL as well as to keep OS community updated with ODL progress
- Doing some coordination among the OpenDaylight projects consuming the Neutron interface so maybe one day we do not need to shutdown all except one
- Updating and maintaining the OpenDaylight Neutron interface and the corresponding driver in OpenStack
On Jan 13, 2015, at 11:16 AM, Kyle Mestery < mestery@...> wrote:
Great sentiment! In a similar manner, I think we should also be sensitive to the people who have been working on this for two years, and are currently driving to fix things from both sides. I'm happy to open this up more, but at the same time, lets not act like no one has been looking at this. People are actively hammering on this, and in fact we've just recently been happy to welcome a plethora of new people to this team.
But I'm curious here: What parts need to bubble up past OVSDB? Because the parts above it are just passing through the Neutron APIs. The actual implementation of how things are programmed on the hosts running the VMs is all done by OVSDB.
|
|
|
Kyle Mestery <mestery@...>
On Tue, Jan 13, 2015 at 11:04 PM, Luis Gomez <ecelgp@...> wrote:
|
|
|
Thank Kyle, as I announced a few days ago, one of the integration priorities this release in Neutron system test so we will be syncing with you and other people involved in Neutron work to come up with a good strategy for it.
BR/Luis
On Jan 14, 2015, at 6:08 AM, Kyle Mestery < mestery@...> wrote:
On Tue, Jan 13, 2015 at 11:04 PM, Luis Gomez <ecelgp@...> wrote:
|
|
|
