[neutron] Drivers meeting agenda for 18.06.2021
Hi, Agenda for tomorrow's drivers meeting is at [1]. We have one new RFE to discuss [2] - It's another approach to do OpenFlow based DVR :) We have also one new RFE [3] which I would like You to check and help triaging before we will discuss it in the drivers meeting. [1] https://wiki.openstack.org/wiki/Meetings/NeutronDrivers#Agenda [2] https://bugs.launchpad.net/neutron/+bug/1931953 [3] https://bugs.launchpad.net/neutron/+bug/1932154 -- Slawek Kaplonski Principal Software Engineer Red Hat
Anybody can help confirm this meeting is at 14:00 UTC Fri, Beijing Time 22:00 UTC+8 Fri? -----邮件原件----- 发件人: Slawek Kaplonski [mailto:skaplons@redhat.com] 发送时间: 2021年6月17日 21:02 收件人: miguel@mlavalle.com; amotoki@gmail.com; YAMAMOTO Takashi <yamamoto@midokura.com>; ralonsoh@redhat.com; Nate Johnson <njohnson@redhat.com>; Haley, Brian <bhaley@redhat.com> 抄送: openstack-discuss@lists.openstack.org 主题: [neutron] Drivers meeting agenda for 18.06.2021 Hi, Agenda for tomorrow's drivers meeting is at [1]. We have one new RFE to discuss [2] - It's another approach to do OpenFlow based DVR :) We have also one new RFE [3] which I would like You to check and help triaging before we will discuss it in the drivers meeting. [1] https://wiki.openstack.org/wiki/Meetings/NeutronDrivers#Agenda [2] https://bugs.launchpad.net/neutron/+bug/1931953 [3] https://bugs.launchpad.net/neutron/+bug/1932154 -- Slawek Kaplonski Principal Software Engineer Red Hat
You can check the time in your local time by clicking the link of the meeting time (1400 UTC) in https://meetings.opendev.org/#Neutron_drivers_Meeting (the neutron drivers meeting info). The URL https://www.timeanddate.com/worldclock/fixedtime.html?hour=14&min=00&sec=0 says 2200 UTC at Beijing. Akihiro Motoki (irc: amotoki) On Fri, Jun 18, 2021 at 10:17 AM Yi Yang (杨燚)-云服务集团 <yangyi01@inspur.com> wrote:
Anybody can help confirm this meeting is at 14:00 UTC Fri, Beijing Time 22:00 UTC+8 Fri?
-----邮件原件----- 发件人: Slawek Kaplonski [mailto:skaplons@redhat.com] 发送时间: 2021年6月17日 21:02 收件人: miguel@mlavalle.com; amotoki@gmail.com; YAMAMOTO Takashi <yamamoto@midokura.com>; ralonsoh@redhat.com; Nate Johnson <njohnson@redhat.com>; Haley, Brian <bhaley@redhat.com> 抄送: openstack-discuss@lists.openstack.org 主题: [neutron] Drivers meeting agenda for 18.06.2021
Hi,
Agenda for tomorrow's drivers meeting is at [1]. We have one new RFE to discuss [2] - It's another approach to do OpenFlow based DVR :) We have also one new RFE [3] which I would like You to check and help triaging before we will discuss it in the drivers meeting.
[1] https://wiki.openstack.org/wiki/Meetings/NeutronDrivers#Agenda [2] https://bugs.launchpad.net/neutron/+bug/1931953 [3] https://bugs.launchpad.net/neutron/+bug/1932154
-- Slawek Kaplonski Principal Software Engineer Red Hat
Akihiro, got it, thank you so much. I also used timeanddate.com to check this. -----邮件原件----- 发件人: Akihiro Motoki [mailto:amotoki@gmail.com] 发送时间: 2021年6月18日 9:44 收件人: Yi Yang (杨燚)-云服务集团 <yangyi01@inspur.com> 抄送: skaplons@redhat.com; miguel@mlavalle.com; yamamoto@midokura.com; ralonsoh@redhat.com; njohnson@redhat.com; bhaley@redhat.com; openstack-discuss@lists.openstack.org 主题: Re: [neutron] Drivers meeting agenda for 18.06.2021 You can check the time in your local time by clicking the link of the meeting time (1400 UTC) in https://meetings.opendev.org/#Neutron_drivers_Meeting (the neutron drivers meeting info). The URL https://www.timeanddate.com/worldclock/fixedtime.html?hour=14&min=00&sec=0 says 2200 UTC at Beijing. Akihiro Motoki (irc: amotoki) On Fri, Jun 18, 2021 at 10:17 AM Yi Yang (杨燚)-云服务集团 <yangyi01@inspur.com> wrote:
Anybody can help confirm this meeting is at 14:00 UTC Fri, Beijing Time 22:00 UTC+8 Fri?
-----邮件原件----- 发件人: Slawek Kaplonski [mailto:skaplons@redhat.com] 发送时间: 2021年6月17日 21:02 收件人: miguel@mlavalle.com; amotoki@gmail.com; YAMAMOTO Takashi <yamamoto@midokura.com>; ralonsoh@redhat.com; Nate Johnson <njohnson@redhat.com>; Haley, Brian <bhaley@redhat.com> 抄送: openstack-discuss@lists.openstack.org 主题: [neutron] Drivers meeting agenda for 18.06.2021
Hi,
Agenda for tomorrow's drivers meeting is at [1]. We have one new RFE to discuss [2] - It's another approach to do OpenFlow based DVR :) We have also one new RFE [3] which I would like You to check and help triaging before we will discuss it in the drivers meeting.
[1] https://wiki.openstack.org/wiki/Meetings/NeutronDrivers#Agenda [2] https://bugs.launchpad.net/neutron/+bug/1931953 [3] https://bugs.launchpad.net/neutron/+bug/1932154
-- Slawek Kaplonski Principal Software Engineer Red Hat
Hi Slawek, I will not be able to attend the meeting tomorrow, have a conflicting appointment, notes on the two RFEs below. On 6/17/21 9:01 AM, Slawek Kaplonski wrote:
Hi,
Agenda for tomorrow's drivers meeting is at [1]. We have one new RFE to discuss [2] - It's another approach to do OpenFlow based DVR :) We have also one new RFE [3] which I would like You to check and help triaging before we will discuss it in the drivers meeting.
[1] https://wiki.openstack.org/wiki/Meetings/NeutronDrivers#Agenda [2] https://bugs.launchpad.net/neutron/+bug/1931953
(Openflow-based DVR L3) As Liu mentioned in the bug, there was a spec/blueprint/almost-implementation before we decided to adopt OVN as the flow-based DVR option. While I can understand the desire to want to have less hops in ML2/OVS/DVR, we made the decision to invest in OVN going forward, so I don't think we should take on this work.
(Off-path SmartNIC Port Binding) From a high-level I'm fine with this, I'm sure Rodolfo will have more questions :) After reading the Nova spec Sean asked a good question - is there a requirement on the core OVN code to support this? I'll add that to the bug (done). -Brian
Hi, Brian Is OVN only one option for Neutron now? The old blueprint is obsoleted because nobody will do this, I don't think it has been almost done. -----邮件原件----- 发件人: Brian Haley [mailto:haleyb.dev@gmail.com] 发送时间: 2021年6月18日 10:16 收件人: Slawek Kaplonski <skaplons@redhat.com>; miguel@mlavalle.com; amotoki@gmail.com; YAMAMOTO Takashi <yamamoto@midokura.com>; ralonsoh@redhat.com; Nate Johnson <njohnson@redhat.com> 抄送: openstack-discuss@lists.openstack.org 主题: [lists.openstack.org代发]Re: [neutron] Drivers meeting agenda for 18.06.2021 Hi Slawek, I will not be able to attend the meeting tomorrow, have a conflicting appointment, notes on the two RFEs below. On 6/17/21 9:01 AM, Slawek Kaplonski wrote:
Hi,
Agenda for tomorrow's drivers meeting is at [1]. We have one new RFE to discuss [2] - It's another approach to do OpenFlow based DVR :) We have also one new RFE [3] which I would like You to check and help triaging before we will discuss it in the drivers meeting.
[1] https://wiki.openstack.org/wiki/Meetings/NeutronDrivers#Agenda [2] https://bugs.launchpad.net/neutron/+bug/1931953
(Openflow-based DVR L3) As Liu mentioned in the bug, there was a spec/blueprint/almost-implementation before we decided to adopt OVN as the flow-based DVR option. While I can understand the desire to want to have less hops in ML2/OVS/DVR, we made the decision to invest in OVN going forward, so I don't think we should take on this work.
(Off-path SmartNIC Port Binding) From a high-level I'm fine with this, I'm sure Rodolfo will have more questions :) After reading the Nova spec Sean asked a good question - is there a requirement on the core OVN code to support this? I'll add that to the bug (done). -Brian
Hi Yi, On 6/17/21 11:27 PM, Yi Yang (杨燚)-云服务集团 wrote:
Hi, Brian
Is OVN only one option for Neutron now? The old blueprint is obsoleted because nobody will do this, I don't think it has been almost done.
I would encourage you to discuss your RFE at the drivers meeting tomorrow, I was just stating my opinion since I will not be there. Part of this is based on the previous spec that was implementing this, https://review.opendev.org/c/openstack/neutron-specs/+/629761 - which was abandoned with this note: "With our continues development of DVR – Openflow solution and the evolution of OVN, we have done some analysis. The feature set in OVN is richer than the current DVR solution & is gaining momentum in the community. So we are intending to transition our DVR efforts to OVN and stop further development in DVR. We believe this will better help community and networking project. Intel will be glad to help customers transition from DVR to OVN." There were a number of patches that were pretty far along when this was done according to what's listed on https://blueprints.launchpad.net/neutron/+spec/openflow-based-dvr -Brian
-----邮件原件----- 发件人: Brian Haley [mailto:haleyb.dev@gmail.com] 发送时间: 2021年6月18日 10:16 收件人: Slawek Kaplonski <skaplons@redhat.com>; miguel@mlavalle.com; amotoki@gmail.com; YAMAMOTO Takashi <yamamoto@midokura.com>; ralonsoh@redhat.com; Nate Johnson <njohnson@redhat.com> 抄送: openstack-discuss@lists.openstack.org 主题: [lists.openstack.org代发]Re: [neutron] Drivers meeting agenda for 18.06.2021
Hi Slawek,
I will not be able to attend the meeting tomorrow, have a conflicting appointment, notes on the two RFEs below.
On 6/17/21 9:01 AM, Slawek Kaplonski wrote:
Hi,
Agenda for tomorrow's drivers meeting is at [1]. We have one new RFE to discuss [2] - It's another approach to do OpenFlow based DVR :) We have also one new RFE [3] which I would like You to check and help triaging before we will discuss it in the drivers meeting.
[1] https://wiki.openstack.org/wiki/Meetings/NeutronDrivers#Agenda [2] https://bugs.launchpad.net/neutron/+bug/1931953
(Openflow-based DVR L3)
As Liu mentioned in the bug, there was a spec/blueprint/almost-implementation before we decided to adopt OVN as the flow-based DVR option. While I can understand the desire to want to have less hops in ML2/OVS/DVR, we made the decision to invest in OVN going forward, so I don't think we should take on this work.
(Off-path SmartNIC Port Binding)
From a high-level I'm fine with this, I'm sure Rodolfo will have more questions :) After reading the Nova spec Sean asked a good question - is there a requirement on the core OVN code to support this? I'll add that to the bug (done).
-Brian
Thanks Brian, this is comments from my previous colleague at Intel, I know that, it isn't Openstack community's decision, per my understanding, it is acceptable if someone can continue to do that, that is to say, this is Intel Openstack team's decision, not openstack community's conclusion. "With our continues development of DVR – Openflow solution and the evolution of OVN, we have done some analysis. The feature set in OVN is richer than the current DVR solution & is gaining momentum in the community. So we are intending to transition our DVR efforts to OVN and stop further development in DVR. We believe this will better help community and networking project. Intel will be glad to help customers transition from DVR to OVN." OVN is an option, but I don't think it is best option unless OVN team can use etcd to replace OVSDB, OVN team is struggling to fix its chicken rib by using DDlog, so far I don't see that Openstack users has big passion to switch to OVN. -----邮件原件----- 发件人: Brian Haley [mailto:haleyb.dev@gmail.com] 发送时间: 2021年6月18日 11:56 收件人: Yi Yang (杨燚)-云服务集团 <yangyi01@inspur.com>; skaplons@redhat.com; miguel@mlavalle.com; amotoki@gmail.com; yamamoto@midokura.com; ralonsoh@redhat.com; njohnson@redhat.com 抄送: openstack-discuss@lists.openstack.org 主题: Re: 答复: [lists.openstack.org代发]Re: [neutron] Drivers meeting agenda for 18.06.2021 Hi Yi, On 6/17/21 11:27 PM, Yi Yang (杨燚)-云服务集团 wrote:
Hi, Brian
Is OVN only one option for Neutron now? The old blueprint is obsoleted because nobody will do this, I don't think it has been almost done.
I would encourage you to discuss your RFE at the drivers meeting tomorrow, I was just stating my opinion since I will not be there. Part of this is based on the previous spec that was implementing this, https://review.opendev.org/c/openstack/neutron-specs/+/629761 - which was abandoned with this note: "With our continues development of DVR – Openflow solution and the evolution of OVN, we have done some analysis. The feature set in OVN is richer than the current DVR solution & is gaining momentum in the community. So we are intending to transition our DVR efforts to OVN and stop further development in DVR. We believe this will better help community and networking project. Intel will be glad to help customers transition from DVR to OVN." There were a number of patches that were pretty far along when this was done according to what's listed on https://blueprints.launchpad.net/neutron/+spec/openflow-based-dvr -Brian
-----邮件原件----- 发件人: Brian Haley [mailto:haleyb.dev@gmail.com] 发送时间: 2021年6月18日 10:16 收件人: Slawek Kaplonski <skaplons@redhat.com>; miguel@mlavalle.com; amotoki@gmail.com; YAMAMOTO Takashi <yamamoto@midokura.com>; ralonsoh@redhat.com; Nate Johnson <njohnson@redhat.com> 抄送: openstack-discuss@lists.openstack.org 主题: [lists.openstack.org代发]Re: [neutron] Drivers meeting agenda for 18.06.2021
Hi Slawek,
I will not be able to attend the meeting tomorrow, have a conflicting appointment, notes on the two RFEs below.
On 6/17/21 9:01 AM, Slawek Kaplonski wrote:
Hi,
Agenda for tomorrow's drivers meeting is at [1]. We have one new RFE to discuss [2] - It's another approach to do OpenFlow based DVR :) We have also one new RFE [3] which I would like You to check and help triaging before we will discuss it in the drivers meeting.
[1] https://wiki.openstack.org/wiki/Meetings/NeutronDrivers#Agenda [2] https://bugs.launchpad.net/neutron/+bug/1931953
(Openflow-based DVR L3)
As Liu mentioned in the bug, there was a spec/blueprint/almost-implementation before we decided to adopt OVN as the flow-based DVR option. While I can understand the desire to want to have less hops in ML2/OVS/DVR, we made the decision to invest in OVN going forward, so I don't think we should take on this work.
(Off-path SmartNIC Port Binding)
From a high-level I'm fine with this, I'm sure Rodolfo will have more questions :) After reading the Nova spec Sean asked a good question - is there a requirement on the core OVN code to support this? I'll add that to the bug (done).
-Brian
participants (4)
-
Akihiro Motoki
-
Brian Haley
-
Slawek Kaplonski
-
Yi Yang (杨燚)-云服务集团