<fnordahl>
Will continue to review to ensure that the pieces that are prerequisites for planned work for OVN 25.03 are in good shape. Will follow up on this further in Monday's A/V community meeting.
<fnordahl>
I raised a revert for workaround in OVS due to bug in Ubuntu kernel in GHA images, as that is now fixed. Thanks for merge imaximets!
<fnordahl>
Other than that we'll soon put a snapshot of OVS into the development release of Debian/Ubuntu to accomodate work on DPDK 24.11.
<fnordahl>
That's it for me.
<imaximets>
fnordahl, thanks!
<imaximets>
Who's next?
<mkalcok>
I can go next
<imaximets>
mkalcok, sure.
<mkalcok>
I've picked up old patchset from Felix that adds IPv4 over IPv6 routing.
<mkalcok>
It was not merged because numans found a issue with register corruption in combination with distributed NAT
<mkalcok>
I attempted to fix it by using different registers for IPV4_SRC and it seems to work fine.
<mkalcok>
I posted patches and fnordahl pointed out that it could use some more tests, so I'll update those
<mkalcok>
and that's all from my side
<numans>
mkalcok++
<imaximets>
Thanks, mkalcok!
<imaximets>
Who wants to go next?
<numans>
I can go real quick. Nothing much from my side. I did a few reviews.
<imaximets>
Is that all? :)
<numans>
Still working on the northd I-P for logical switch create/delete.
<numans>
That's it now from me :)
<imaximets>
ack.
<imaximets>
Thanks, numans!
<imaximets>
I guess, I o have a small update
<imaximets>
*do
<imaximets>
I've been working on enabling prefix matches by default in OVS for IPv6 addresses to avoid datapath flow explosion.
<imaximets>
Will look into enabling these prefix matches from OVN side.
<imaximets>
Since they will only be enabled by default in 3.5+
<imaximets>
I also looked at reducing IPv6 matches caused by IPv6 prefix delegation flows.
<imaximets>
Sent some patches to not generate these logical flows when prefix delegation is not configured.
<imaximets>
These are also now merged, thanks to dceara !
<imaximets>
And that's it from my side.
<imaximets>
Who wants to go next?
<amusil>
I can
<imaximets>
amusil, go ahead.
<amusil>
Posted v2 RFC for the commit all patch, and one patch that we will need to enable that from OvS side to bump the number of recirculations in userspace dp
<amusil>
There was one bug with lb_force_snat_ip when there is VIP that is the same as client sending packets towards the backend
<fnordahl>
as in bumping the limits? with imaximets presentation fresh in mind, I hope we have enough stack space! :)
<imaximets>
Yeah, I'm a little concerned about that one.
<amusil>
Posted a patch for that, other than that I've been focusing mostly on the transit router, hoping to post the series tomorrow
<amusil>
That's it, thanks
<amusil>
The second option is to disable those tests for userspace DP
<imaximets>
Thanks, amusil ! Let's discuss the recirculation limit on the list.
<amusil>
Ack, thanks
<imaximets>
Anyone else wants to share?
<mj2>
sure i can go
<imaximets>
mj2, sure. Go right ahead!
<mj2>
i have mainly been working on getting LRPs working without ipv4
<mj2>
i am also working on a patch for allowing RAs to be sent without any ipv6 prefixes
<mj2>
as only ipv6 lla is needed
<mj2>
currently adding more testing to the latter and working on documentation on the former
<mj2>
but its steady progress
<mj2>
:+1:
<imaximets>
Nice! Thanks, mj2 ! Anything else?
<mj2>
nope
<imaximets>
OK.
<imaximets>
Do we have anyone else here today?
<imaximets>
I'll take that as a 'no'. :)
<imaximets>
If we don't have anything else to discuss, let's call it a meeting.
<mkalcok>
thanks all \o
<imaximets>
Thanks, everyone!
<fnordahl>
\o cheers and thanks!
<imaximets>
Bye.
<mj2>
c ya
<mj2>
:+1:
<amusil>
Thanks, bye
mkalcok has quit [Quit: leaving]
amusil has quit [Quit: Client closed]
mj2 has left #openvswitch [WeeChat 4.1.1]
<_lore_>
bye
<numans>
bye
kuraudo has quit [Remote host closed the connection]