16:19 <imaximets> Morning. :)

16:19 <fnordahl> o/

16:19 <dceara> Hi!

16:19 <imaximets> I think Mark is away today, so maybe I'll host the OVN meeting.

16:19 <dceara> imaximets++

16:20 <amusil> Hi

16:20 <imaximets> I don't have much from my side, been stuck debugging ipsec for the past two weeks, so who wants to start?

16:20 <dceara> I don't have much either so I can go next. :)

16:21 <imaximets> dceara, sure.

16:21 <dceara> One upstream related thing was that hopefully our CI jobs that run ovn-kubernetes tests might be a bit more stable from now on.

16:22 <dceara> ovn-kubernetes folks just merged a backport PR to their release-1.0 branch that should make it less flaky.

16:22 <dceara> I'm sure there are more issues but we're planning on addressing those in the future by trying to consolidate the CI effort between the two projects.

16:22 <dceara> Another thing I wanted to bring up was a question to fnordahl.

16:23 <fnordahl> dceara: go for it

16:23 <dceara> Related to the BGP effort, I was wondering if you had some time to outline the tasks you were planning on working on in the next dev cycle.

16:23 <dceara> We (Red Hat) will probably want to build on top of that too.

16:24 <dceara> That's it on my side, thanks.

16:24 <fnordahl> I'll take that as cue for my update :) Mostly here to say hello, my non-update is that I'm in crunch for delivering on downstream roadmap commitments for another week. Will try to put together that doc for coming cycles upstream bgp related work as promised asap.

16:24 <_lore_> hi all

16:25 <fnordahl> Two weeks after next is planning anway, so that will follow naturally

16:25 <dceara> fnordahl, great, thanks!

16:25 <fnordahl> But will expedite that specific doc

16:25 <fnordahl> That's it fo rme

16:26 <imaximets> fnordahl, would be good if you - as a sole decision maker for what is included into Ubuntu kernel :) - checked on OVS fixes that are still not there.

16:26 <fnordahl> lol, I saw that pointed ovs-dev subject fly by, I'll look into it

16:26 <fnordahl> imaximets: thanks!

16:26 <imaximets> :D

16:27 <imaximets> fnordahl, thanks!

16:27 <imaximets> Who wants to go next?

16:27 <_lore_> I can go next

16:27 <_lore_> quite fast

16:27 <imaximets> _lore_, go ahead!

16:27 <_lore_> I have mainly working on ovn sampling issue, posted a fix for it

16:28 <_lore_> then I would like to help the bgp effort :) so I will wait for fome feedbacks from fnordahl

16:28 <_lore_> that's all from my side

16:29 <imaximets> Thnaks, _lore_ !

16:29 <imaximets> *thanks

16:29 <imaximets> Who's next?

16:29 <amusil> I can go quickly

16:29 <imaximets> amusil, sure

16:30 <amusil> I spent most of the time looking through the commit all for LR, it would be nice to get some feedback on the RFC https://patchwork.ozlabs.org/project/ovn/patch/20241008101155.331151-3-amusil@redhat.com/

16:30 <amusil> zhouhan If you have some time please take a look

16:30 <zhouhan> amusil: ack

16:31 <amusil> Other than that some smaller things, that's it, thanks

16:31 <imaximets> amusil, thanks!

16:31 <zhouhan> I can go quickly

16:31 <zhouhan> I am working on an RAFT schema upgrade scale issue - the memory spike on the server. Got some pointers from imaximets last week.

16:31 <imaximets> zhouhan, go ahead!

16:32 <imaximets> ack. Any preliminary results?

16:32 <zhouhan> The findings are: 1. the initial update is a reply to client request, so the backpressure doesn't apply 2. the buffer was actually sent to socket, so backpressure is also not easy (if possible at all)

16:33 <zhouhan> It seems difficult to implement a ideal solution for this.

16:33 <imaximets> Makes sense.

16:34 <imaximets> That aligns with what I've seen in the field as well.

16:34 <zhouhan> I am currently looking into a workaround to take care of our downstream operation for now, such as adding some naive delay when sending too much data in one loop

16:34 <zhouhan> but definitely not something formal

16:34 <zhouhan> that's it from me

16:35 <zhouhan> (any other suggestions would be very welcome)

16:35 <imaximets> OK. I'll try to think about this on the background. But yeah, it's not a simple issue.

16:35 <imaximets> Thanks, zhouhan !

16:35 <imaximets> Anyone else here wants to share?

16:36 <imaximets> I'll take silence as a 'no'.

16:36 <imaximets> Have a nice rest of your day, everyone!

16:36 <fnordahl> \o cheers!

16:37 <imaximets> Bye!

16:37 <dceara> Thanks, bye!

16:37 <zhouhan> thanks, bye!

16:37 <_lore_> bye

16:38 <amusil> Thanks, bye