14:00:10 #startmeeting neutron_drivers 14:00:10 Meeting started Fri Sep 9 14:00:10 2022 UTC and is due to finish in 60 minutes. The chair is lajoskatona. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:10 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:10 The meeting name has been set to 'neutron_drivers' 14:00:12 o/ 14:00:12 o/ 14:00:16 hi 14:00:26 hi 14:00:43 hi 14:00:53 o/ 14:02:06 Let's start we have quoume as I count 14:02:14 We have one RFE for today 14:02:17 [RFE] OVN support for vnic type virtio-forwarder (#link https://bugs.launchpad.net/neutron/+bug/1988542 ) 14:02:34 Hi, this RFE is reported by me 14:03:04 If I understand well this is realted to some smart-nic feature, am I rigth? 14:03:16 Hi skingFD, thanks for proposing 14:03:39 Yes, it is a new feature related to smart-nic 14:04:10 It is a standard already supported by NOVA and some smartnics, while neutron does not 14:04:28 the nova spec speaks about netronome only, and as I remember some work was done in Neutron also 14:04:32 for OVS 14:04:47 Yes, the ovs plugin already supports it 14:04:56 but the ovn still does not 14:04:57 are there any more changes required on neutron side for that, besides what You already proposed? 14:05:34 I don't think there are more changes needed 14:05:56 then it seems like a no brainer to me 14:05:58 It is a open-sourced vnic type, beside netronome, everyone can use it 14:06:02 Agree 14:06:19 cool then, +1 from me as well 14:06:25 even an RFE is too much perhaps, but we discussed it and have clear picture 14:06:29 +1 14:07:13 thx 14:07:24 +1 14:07:28 +1 14:07:37 to formalize my position.... +1 14:07:45 We have agreement than :-) 14:08:00 I like this type of RFE 14:08:05 thx 14:08:05 skingFD: thanks for proposing and working on this 14:08:09 straightforward 14:08:13 yeah 14:08:17 given the evidence 14:08:42 #topic On Demand Agenda 14:08:53 Do you have anything to discuss today? 14:08:54 Appreciate more code reviewing on the patch I've supported 14:09:09 It's my first time contributing to neutron 14:09:09 skingFD: I'll take a look soon 14:09:28 and thank you for your contribution. we welcome you! 14:09:31 So more check is needed I guess :) 14:09:35 skingFD: sure, we are now closing the Zed release so next weeks will be busy 14:09:58 :mlavalle Thanks! 14:10:51 Do we have time for another small RFE? 14:11:13 ges: I think yes 14:11:26 we have plenty of time still :) 14:11:31 [rfe] rate-limit metadata API : https://bugs.launchpad.net/neutron/+bug/1989199 14:11:38 keep in time haleyb will drop off soon 14:11:44 will we still have quorum? 14:12:02 mlavalle: i'm still here for now :) 14:12:16 ok 14:12:35 LOL, this RFE is hot just out of the oven 14:12:41 yep! 14:12:55 be careful, don't burn yourselves 14:13:24 :-) 14:13:59 too early maybe? I understand if we need to wait a bit! 14:14:28 naah, I don't think so 14:14:41 ges: do You want to do it with config option so it will be for all networks or routers on nodes? 14:14:41 ges: is this for ml2/ovs 14:14:43 ? 14:14:50 or somehow through API per network or router? 14:15:15 mlavalle: I think that the same can apply to ovn as well 14:15:20 slaweq: in our PoC I have done that with some config options 14:15:21 I know some users set rate limiting for metadata at the VIP level, having this native does make sense to me 14:15:22 agree 14:15:27 as we do have haproxy in the ovn-meta namespace there 14:16:06 the rate-limiting would be done by source ip 14:16:39 https://www.haproxy.com/blog/four-examples-of-haproxy-rate-limiting/#sliding-window-rate-limiting 14:17:02 I think it might worth a spec to discuss options 14:17:11 +1 14:17:34 mlavalle: I'm not sure I understand how it relates to either ml2 or ovs :/ 14:17:37 I agree 14:18:14 for example: will it be just config knob to enable/disable rate-limit or will there be knob to configure actual limit of requests 14:18:28 ges: In neutron there's 2 metadata agent now one for OVN and one for all other (OVS, linuxbridge) 14:19:44 i would think we could enable this by default, if it's by source IP it should be Ok and not impact normal operations, correct? 14:20:34 haleyb: as long as it will be set to some reasonable value to not affect e.g. cloud-init during boot of the vms 14:20:46 lajoskatona: ah I see! We don't use OVN but I suppose it could go there as well 14:20:57 ges: yes, exactly 14:21:20 As I see there's a lot of things which should be clarified in a spec 14:21:58 So to summarize: the RFE generally looks ok, but we need a spec to discuss the details, is that correct? 14:22:12 +1 14:22:16 slaweq: yes, making it large enough to cover the cloud-init case +some% should limit someone doing a DoS after they boot 14:22:44 +1 for me 14:22:47 Ok, well my repair guy is here, have to run, i'm fine with this for now... 14:22:48 +1 14:22:50 +1 14:22:58 haleyb: thanks 14:23:11 thanks! 14:23:40 Ok, I wil update the RFE, thanks ges for proposing it 14:24:13 yw! next step is to write a spec, right? Should I do it? 14:25:00 ges: yes, here is a template: https://opendev.org/openstack/neutron-specs/src/branch/master/specs/zed/placeholder.rst 14:25:16 soon there will be a new folder for Antelope/2023.1 14:25:34 I think the folder is already there 14:26:00 and it is really there 14:26:01 https://opendev.org/openstack/neutron-specs/src/branch/master/specs/2023.1 14:26:13 thanks mlavalle 14:26:20 yeah, I recently approved the patch ;-) 14:26:37 Alright! 14:26:38 I looked for the end of alphabet as usual, and we reached the end of it :-) 14:26:53 LOL, you just went to the end 14:26:53 Is there anything we can discuss today? 14:28:07 nothing from me 14:28:14 neither from me 14:28:18 If nothing more , let's close the meeting for today 14:28:24 Thanks for coming :-) 14:28:29 #endmeeting