14:00:26 <mlavalle> #startmeeting neutron_drivers
14:00:27 <openstack> Meeting started Fri Jul 19 14:00:26 2019 UTC and is due to finish in 60 minutes.  The chair is mlavalle. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:29 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:31 <openstack> The meeting name has been set to 'neutron_drivers'
14:00:33 <njohnston> o/
14:00:37 <ralonsoh> hi
14:00:43 <adrianc> hello
14:01:05 <mlavalle> let's wait a couple of minutes
14:01:26 <mlavalle> haleyb will not join us today. dentist appointment.... ouch!!!!
14:01:41 <slaweq> hi
14:02:59 <amotoki> hi
14:03:46 <mlavalle> ok, we have quorum, so let's get rolling
14:03:52 <mlavalle> #topic RFEs
14:04:22 <mlavalle> For today we will re-visit adrianc's proposal: https://bugs.launchpad.net/neutron/+bug/1834176
14:04:23 <openstack> Launchpad bug 1834176 in neutron "[RFE] Neutron enhancements to support per-physnet and IPoIB interface drivers" [Wishlist,In progress] - Assigned to Adrian Chiris (adrian.chiris)
14:04:45 <mlavalle> he repurposed it based on last week's feedback. Thanks!
14:04:54 <adrianc> np :)
14:08:52 <njohnston> so it makes sense to me
14:09:35 <slaweq> if this will go to networking-mlnx repo, than I'm totally fine with this too :)
14:10:12 <adrianc> by this you mean the ipoib and per-physnet interface drivers ?
14:10:16 <njohnston> I think the ask for us is for the interface he defines in bullet points 1-3 in the RFE in order to enable networking-mlnx to accomplish this
14:10:17 <mlavalle> yeah, looked quickly over the changes and they all seem pretty sensible
14:10:52 <amotoki> what is the relationship between interface driver and vif_type?
14:11:03 <adrianc> the idea was to keep things to a minimum, in order to allow the per physnet and ipoib drivers to exist in networking-mlnx
14:11:26 <adrianc> amotoki: what do you mean ?
14:11:51 <amotoki> adrianc: I wonder we really need to pass an intreface driver over RPC.
14:12:14 <amotoki> adrianc: which interface driver is used depends on mechanism driver.
14:12:32 <adrianc> interface driver requires knowledge about the network - e.g to know which physnet it belogns to to invoke the correct interface driver
14:12:44 <adrianc> i mean the per-physnet interface driver
14:13:09 <adrianc> an the ipoib one requires knowledge of the network's segmentation ID
14:13:16 <adrianc> due to iproute2 limitations
14:13:38 <adrianc> i.e to create that interface you require to specify that
14:13:43 <slaweq> can we maybe add segmentation_id to the port send to dhcp/l3 agent?
14:13:53 <slaweq> instead of doing new rpc calls to server
14:13:57 <amotoki> passing segmentation ID looks okay
14:14:00 <slaweq> (I'm just thinking loud now)
14:14:38 <adrianc> the callback for the interface driver should be consistent, remember both dhcp and l3 agents use it
14:17:18 <mlavalle> I'd say we can approve this RFE and discuss the details in gerrit
14:17:25 <amotoki> it seems I am talking about impl detai. what the RFE request in general totally makes sense.
14:17:37 <amotoki> so +1 to approve this RFE
14:18:26 <slaweq> +1
14:18:28 <njohnston> +1
14:18:42 <mlavalle> ok, approved it is
14:18:56 <amotoki> let me note I am not confident whether kwargs or more explicit definition is better as interface driver interface
14:19:09 <amotoki> let's defer it to the review.
14:19:22 <mlavalle> adrianc: thanks for the submission and please go and enjoy the rest of your weekend
14:19:41 <mlavalle> for today, I only had one RFE
14:19:42 <adrianc> thank you all :)
14:20:00 <mlavalle> is there anything else we should discuss today?
14:21:27 <mlavalle> ok, thanks for attending
14:21:34 <mlavalle> have a great weekend!
14:21:35 <njohnston> o/
14:21:40 <ralonsoh> bye
14:21:50 <amotoki> have a nice weekend,!
14:22:03 <slaweq> bye
14:22:07 <mlavalle> #endmeeting