Bossi has quit [Quit: No Ping reply in 180 seconds.]
Bossi has joined #openvswitch
ChmEarl has quit [Quit: Leaving]
mtomaska_ has joined #openvswitch
mtomaska has quit [Ping timeout: 244 seconds]
_lore_ has quit [*.net *.split]
_lore_ has joined #openvswitch
kuraudo has joined #openvswitch
froyo has joined #openvswitch
vlrpl has joined #openvswitch
amorenoz has joined #openvswitch
kuraudo has quit [Ping timeout: 244 seconds]
kuraudo has joined #openvswitch
kuraudo has quit [Client Quit]
kuraudo has joined #openvswitch
kuraudo has quit [Ping timeout: 260 seconds]
GNUmoon has quit [Ping timeout: 264 seconds]
GNUmoon has joined #openvswitch
kuraudo has joined #openvswitch
kuraudo has quit [Quit: kuraudo]
kuraudo has joined #openvswitch
amorenoz has quit [Remote host closed the connection]
kuraudo has quit [Ping timeout: 276 seconds]
kuraudo has joined #openvswitch
dceara has joined #openvswitch
kuraudo has quit [Quit: kuraudo]
kuraudo has joined #openvswitch
kuraudo has quit [Ping timeout: 248 seconds]
ChmEarl has joined #openvswitch
kuraudo has joined #openvswitch
kuraudo has quit [Ping timeout: 252 seconds]
roriorden has joined #openvswitch
roriorden_ has joined #openvswitch
mmichelson has joined #openvswitch
roriorden has quit [Ping timeout: 276 seconds]
<mmichelson>
Hi everyone, it's time to begin the OVN weekly developers' meeting.
<fnordahl>
o/
<mmichelson>
I can get things started. I have worked on the new "allow-established" ACL patch, and have been working on the feedback I received from amusil. Other than that, it's been internal Red Hat stuff.
<mmichelson>
That's all from me.
<mmichelson>
I'd like to get to reviews soon, because otherwise January is going to be a rough month.
<mmichelson>
Who'd like to go next?
<imaximets>
o/
<imaximets>
May I?
<mmichelson>
go right ahead imaximets
zhouhan has joined #openvswitch
mkalcok has joined #openvswitch
<imaximets>
I didn't spend too much time on OVN, but we now have STT deprecated in both OVS and OVN.
<mmichelson>
bye bye STT
<imaximets>
Thanks, dceara for reviews!
<imaximets>
I also posted a patch set for OVS to add explicit TLSv1.3
<imaximets>
And some more TLS/OpenSSL cleanups.
<fnordahl>
nice
<imaximets>
Once this is meged in OVS, we'll need to bring those change to OVN
<imaximets>
via a submodule bump + some code updates.
<imaximets>
fnordahl, for the BGP OVS set
<imaximets>
Eelco is almost done with review, he'll post it soon.
<imaximets>
If already didn't
<imaximets>
That's all from my side.
<mkalcok>
(sorry for late join o/)
<mmichelson>
Thanks imaximets. Who's next?
<fnordahl>
imaximets: much apprechiated, thanks!
<fnordahl>
I have an update
<fnordahl>
Other than the before mentioned OVS route-table patches, I see dmarchand has done some work towards the issue we had with DPDK 24.11 memory consumption, so will put that into the environment where we discovered it to help test.
<fnordahl>
Other than that I'm preparing some patches to improve Python handling in the upstream OVS debian sources and will post those shortly.
<fnordahl>
That's it for me.
roriorden_ has quit [Ping timeout: 244 seconds]
<mmichelson>
Thanks fnordahl
<mmichelson>
Who wants to give the next update?
<_lore_>
Can I go next?
<mmichelson>
go right ahead _lore_
<_lore_>
I have a question for zhouhan, I was looking to an issue regarding ovn-ic + vxlan, in facts ovn-ic does not work with vxlan tunneling since ts datapath key does not fit in 12 bit
<_lore_>
it is something like 0xff000x
<_lore_>
I verified traffic is flowing if I let the system select a dynamic datapath key for the transit switch
<_lore_>
so I was wondering if there is a particular reason for that or not
<_lore_>
zhouhan: ..^
<zhouhan>
_lore_, sorry let me check
<mmichelson>
While zhouhan is checking that, did you have anything else you wanted to bring up in your update _lore_ ?
<_lore_>
zhouhan: in allocate_ts_dp_key() you can select an hint + min value
<_lore_>
mmichelson: nope
<zhouhan>
_lore_, can't remember the details now but I recall that dceara had done something related to the vxlan v.s. ovn-ic.
<zhouhan>
_lore_, can you post to ML and I will check offline and reply
<_lore_>
the issue is we have 12 bit for metadata in the vxlan tunnel
<_lore_>
12 bit for outport and 12 bit for meta
<dceara>
zhouhan, _lore_, I had looked into using vxlan with ovn-ic in ovn-kubernetes and realized it's not supported. But didn't do anything to make it work. :)
<mmichelson>
We had an internal discussion about this today, and it seemed that the problem was that "global" tunnel keys were not taking into account the reduced space required by VXLAN,and that just needed to be taken into account.
<zhouhan>
dceara, ok, my bad memory
<_lore_>
I can post an rfc patch and continue the discussion upstream
<_lore_>
thx
<_lore_>
*a
<mmichelson>
The worry was if VXLAN and GENEVE were mixed together, because then there could be questions about whether a key is global or local since the ranges could overlap.
<_lore_>
mmichelson: I guess we can just let ovn select the first available datapath key for the ts
<_lore_>
*selects
<mmichelson>
OK thanks _lore_ for the update. It's probably a good idea to continue the discussion on the mailing list like zhouhan suggested, plus maybe experimenting with your own patches to see how things are affected.
<mmichelson>
In the meantime, does anyone else want to give an update?
<dceara>
I can go quickly.
<dceara>
I applied and backported (when applicable) some of the patches that were in the patchwork queue. I didn't manage to go through the whole list though.
<dceara>
Together with Felix and mkalcok we discussed on the proposed SB schema change for Routes (for BGP integration).
<dceara>
I was about to apply the patch but realized we might need a small change. Waiting for feedback before applying it.
<dceara>
Next I want to finish reviewing Felix's v2 prerequisite series and get those merged hopefully early next week.
<dceara>
It came from a rather long internal discussion but it worries me a bit that it tries to change behavior (and might need another knob - hopefully we can avoid that). In any case some extra eyes on it would be great.
<dceara>
That's it from my side, thanks.
<mmichelson>
Thanks dceara
tpires has joined #openvswitch
<mmichelson>
Does anyone else want to go?
<mmichelson>
I think that's a full minute with no response. Thanks everyone, and have a good day.