felixhuettner has quit [Ping timeout: 252 seconds]
felixhuettner has joined #openvswitch
elvira1 has quit [Quit: WeeChat 4.5.1]
dceara has joined #openvswitch
kuraudo has quit [Remote host closed the connection]
kuraudo has joined #openvswitch
dceara_ has joined #openvswitch
mmichelson has joined #openvswitch
fnordahl has joined #openvswitch
mkalcok has joined #openvswitch
<mmichelson>
Hi everyone, it's time to begin the weekly OVN developers' meeting.
<imaximets>
o/
<felixhuettner>
o/
<mkalcok>
o/
<fnordahl>
o/
<mmichelson>
My update this week is pretty quick. Upstream it's been mostly reviews.
<mmichelson>
I plan to give amusil's next version of the "commit-all" patch for logical routers a look today.
<mmichelson>
I will send branching patches out tomorrow.
<mmichelson>
That's all from me. Who is next?
<imaximets>
May I?
<mmichelson>
go for it imaximets
<imaximets>
Don't have much. Did some reviews as well and posted a set of fixes for the spine-leaf patch set to turn on FDB learning and stop flood in the spine switch.
<imaximets>
That's all from me.
<felixhuettner>
If its ok i'd continue
<mmichelson>
felixhuettner, sure thing. Go right ahead
<felixhuettner>
I first want to take the opportunity and again thank everyone who was involved in the whole BGP series.
<felixhuettner>
I spend most of my time working on smaller fixes (that are mostly merged) and reviewing some other changes.
<felixhuettner>
I also started working on incremental processing changes to learned routes.
<felixhuettner>
Since we don't have any incremental handling for routes at the moment (at least i did not find any) this will get fun.
<felixhuettner>
Thats it from my side
<mmichelson>
Thanks felixhuettner
<mmichelson>
Who wants to go next?
<mkalcok>
I can follow up
<mmichelson>
mkalcok, go right ahead
<mkalcok>
The LB/NAT advertisement is up for review. That was pretty much my only focus. Thanks for the reviews and help so far dceara felixhuettner
amusil has joined #openvswitch
<mkalcok>
that's all from my side
<mmichelson>
thanks mkalcok
<mmichelson>
Who wants to give the next update?
<dceara_>
I can go next, sorry for joining late.
<mmichelson>
dceara_, sure thing
<dceara_>
I've been doing release related reviews last week.
<dceara_>
And I guess we'd also really like to get Ales' [ovs-dev,v5] northd: Add LR option to commit all traffic. in but we're waiting for reviews and test results for zhouhan.
<dceara_>
That's it on my side, thanks!
<mmichelson>
dceara_, I'll be giving my review to that patch later today.
<mmichelson>
Thankd dceara_ . Who wants to go next?
<fnordahl>
I have an update
<mmichelson>
fnordahl, okie-dokie
<fnordahl>
I've mostly been in the background this week due to an internal Mid-Cycle conference, helping out with what I can for the ongoing work from our side.
<fnordahl>
The multi node test is still in progress and we will post that when MJ is ready.
<fnordahl>
Owe a reply to dceara on the question of additional needs for locating instances wrt. announcements, but have been waiting for the above work to use as grounds for discussion. Will be follow-up for next cycle at this point for sure :)
<fnordahl>
One last thing I wondered about is whether the discussion on communication channel went anywhere, the Ubuntu development teams are moving over to Matrix, so just wanted to air the desire to do something other than IRC again :)
<mmichelson>
fnordahl, to my knowledge, the discussion on communication channel never really went anywhere.
<dceara_>
Maybe it's something we should discuss with a wider audience? E.g., on list?
<fnordahl>
Yeah, and after release for sure
<dceara_>
I wonder if there are other contributors that don't join the IRC meeting who would feel more comfortable with joining the weekly meetings on Matrix.
<mmichelson>
If we didn't just have a community meeting on Monday, I'd suggest discussing it in the next one.
<mmichelson>
But that's several weeks from now.
<mmichelson>
So yeah, I think the list is a good idea.
<fnordahl>
I can suggest a agenda item for next A/V meeting
<imaximets>
IMO, the entrance barrier for matrix is either the same or higher.
<dceara_>
We can also move the A/V meeting earlier if people wish. I just counted 6 weeks from now but we can move it earlier.
<mmichelson>
I guess if we have a lot of agenda items for an A/V meeting, we could move it earlier.
<mmichelson>
Or we could even have a special session just for this particular topic.
<mmichelson>
Since I imagine we could talk for a long time about it :)
<fnordahl>
No particular rush on my end, I'll add it to the agenda and we'll see where it goes
<mkalcok>
I have no strong opinion on the A/V meet date, but I will be off during the next proposed session.
<mmichelson>
Anyway, I think for now, let's move that discussion to the list, and if we want to have another A/V meeting sooner, then we can determine that on-list as well.
<mmichelson>
fnordahl, did you have anything else for your update?
<fnordahl>
Oh, sorry, that's it for me
<mmichelson>
fnordahl, OK thanks.
<mmichelson>
Who wants to go next?
<amusil>
I can quickly
<mmichelson>
amusil, sure thing
<amusil>
I spent some time on various reviews and the commit-all patch, that's about it, thanks
<amusil>
The v5 is up and ready on ML
ChmEarl has joined #openvswitch
<zhouhan>
thanks amusil, I will review it today
<mmichelson>
amusil, thanks, I'll have a look at it later as well.
<mmichelson>
OK, who's next?
<zhouhan>
I can go next
<mmichelson>
go right ahead zhouhan
<zhouhan>
I will add an option for the lower priority src-route change, which I think is the most convenient way to proceed
<zhouhan>
That's it from me
<mmichelson>
zhouhan, OK thanks.
<mmichelson>
Does anyone else have anything to bring up?
<mmichelson>
OK, I think that's a "no". zhouhan were you able to get any performance numbers on the commit-all patch?
<zhouhan>
mmichelson: not yet, I am pinging Alin today
<mmichelson>
zhouhan, OK
<zhouhan>
mmichelson: he did test v3 but it didn't work. He started on v4, and hopefully, v5 would be even better
<mmichelson>
zhouhan, ack.
<mmichelson>
I think that's everyone's updates. Thank you everyone, and have a good day.
<dceara_>
zhouhan, just to be sure, the tests Alin ran with v3 were with hwol, right?
<dceara_>
zhouhan, so the regression he reported was from "very high throughput with hwol" to "very low throughput with hwol", correct?
<zhouhan>
dceara_: yes, hwol
<dceara_>
mmichelson, sorry, I thought I'll take the opportunity to ask here while zhouhan is around
<zhouhan>
dceara_: and yes, because hwol didn't work
<dceara_>
zhouhan, OK, so we're not sure about the performance hit with commit-all enabled in non-hwol scenarios.
<zhouhan>
dceara_: correct. Hopefully we should know with v4/v5
<dceara_>
Our internal tests showed 10% decrease IIRC on some specific traffic patterns.
<dceara_>
with v3
<amusil>
v5 has the double register optimization so it should be slightly better than v4
<zhouhan>
amusil: ack
<dceara_>
zhouhan, ok, thanks for the info, now I'm done. :)
<zhouhan>
dceara_: np
<mmichelson>
All right, bye everyone!
mmichelson has quit [Quit: Leaving]
<dceara_>
Thanks, bye!
<mkalcok>
thanks all \o
<zhouhan>
thanks, bye
<imaximets>
Thanks! Bye.
mkalcok has quit [Quit: leaving]
<fnordahl>
\o thanks all
<felixhuettner>
thanks a lot, bye
<amusil>
Thanks, bye
amusil has quit [Quit: Client closed]
dceara_ has quit [Quit: Leaving]
dceara has quit [Ping timeout: 252 seconds]
mj2 has joined #openvswitch
mj2 has quit [Ping timeout: 248 seconds]
kuraudo has quit [Remote host closed the connection]
zhouhan has quit [Quit: Client closed]
mj2 has joined #openvswitch
dceara has joined #openvswitch
dceara has quit [Client Quit]
dceara has joined #openvswitch
mj2 has quit [Ping timeout: 260 seconds]
tpires has joined #openvswitch
tpires_ has quit [Ping timeout: 246 seconds]
otherwiseguy has joined #openvswitch
dceara has quit [Remote host closed the connection]