15:00:14 <ralonsoh> #startmeeting neutron_qos
15:00:15 <openstack> Meeting started Tue Nov 19 15:00:14 2019 UTC and is due to finish in 60 minutes.  The chair is ralonsoh. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:16 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:19 <openstack> The meeting name has been set to 'neutron_qos'
15:00:20 <ralonsoh> Hello
15:00:28 <ralonsoh> today David is not going to attend
15:00:40 <ralonsoh> and the agenda is quite short
15:00:52 <ralonsoh> so let's wait 1 min first
15:01:41 <ralonsoh> #topic RFEs
15:01:48 <ralonsoh> #link https://bugs.launchpad.net/neutron/+bug/1476527
15:01:48 <openstack> Launchpad bug 1476527 in neutron "[RFE] Add common classifier resource" [Wishlist,Triaged] - Assigned to Igor D.C. (igordcard)
15:02:01 <ralonsoh> As commented, David is not going to attend today
15:02:01 <slaweq> hi again
15:02:06 <ralonsoh> hi slaweq
15:02:26 <ralonsoh> David will present the spec in drivers meeting, in Friday
15:02:35 <ralonsoh> as agreed last meeting
15:03:00 <ralonsoh> then, once the spec is approved we can continue with the code review
15:03:10 <slaweq> ralonsoh: please keep in mind that I cancelled this week's drivers meeting
15:03:29 <ralonsoh> yes....
15:03:32 <ralonsoh> hehehehe
15:03:41 <ralonsoh> I'll talk to David then
15:03:59 <slaweq> thx
15:03:59 <ralonsoh> my bad. But he needs to present the spec and RFE to someone else
15:04:11 <slaweq> sorry for that but we will not have quorum this week
15:04:50 <slaweq> ralonsoh: but this rfe https://bugs.launchpad.net/neutron/+bug/1476527 is approved already
15:04:50 <openstack> Launchpad bug 1476527 in neutron "[RFE] Add common classifier resource" [Wishlist,Triaged] - Assigned to Igor D.C. (igordcard)
15:04:56 <ralonsoh> I know, he is trying to push it for the last 18 months
15:05:04 <ralonsoh> I know, but the spec
15:05:12 <ralonsoh> just to know how to implement it
15:05:13 <slaweq> so it's "only" about reviewing spec, right?
15:05:19 <ralonsoh> yes yes
15:06:25 <ralonsoh> ok then, we can move to next section
15:06:35 <ralonsoh> #topic Bugs
15:06:39 <slaweq> it is this spec https://review.opendev.org/#/c/678865/ right?
15:06:55 <ralonsoh> exactly
15:07:05 <ralonsoh> let me send the current links for it
15:07:11 <ralonsoh> #link https://review.opendev.org/678865 : Spec dscp + classifier
15:07:12 <ralonsoh> #link https://review.opendev.org/670049 : Neutron-Lib - classifier migration
15:07:12 <ralonsoh> #link https://review.opendev.org/670050 : Neutron Classifier migration into Neutron
15:07:12 <ralonsoh> #link https://review.opendev.org/636330 : DSCP + Classifier neutron-lib
15:07:12 <ralonsoh> #link https://review.opendev.org/636333 : DSCP + Classifier neutron
15:07:21 <ralonsoh> (all in https://etherpad.openstack.org/p/neutron_qos_meeting_chair)
15:07:26 <slaweq> ok, I will try review it this week
15:08:02 <ralonsoh> slaweq, thanks a lot
15:08:33 <ralonsoh> let's move to bugs
15:08:41 <ralonsoh> #link https://bugs.launchpad.net/neutron/+bug/1845176
15:08:41 <openstack> Launchpad bug 1845176 in neutron "Removing of QoS queue in neutron-ovs-agent fails due to existing references" [Medium,In progress] - Assigned to Rodolfo Alonso (rodolfo-alonso-hernandez)
15:08:45 <ralonsoh> and the patch
15:09:05 <ralonsoh> #link https://review.opendev.org/#/c/687922/
15:09:39 <ralonsoh> I still need to address some problems with fullstack tests
15:09:47 <ralonsoh> https://7513009f5bff8f76e461-f83d06667d580e000031601b82c71a43.ssl.cf2.rackcdn.com/687922/9/check/neutron-fullstack/acf81f5/testr_results.html.gz
15:10:07 <ralonsoh> related to the need cache variables added, of course
15:10:39 <ralonsoh> but at least I think the patch is addressing the problem we have
15:10:57 <ralonsoh> any comment?
15:11:44 <ralonsoh> next one
15:11:56 <ralonsoh> #link https://bugs.launchpad.net/neutron/+bug/1845161
15:11:56 <openstack> Launchpad bug 1845161 in neutron "Neutron QoS Policy lost on interfaces" [High,In progress] - Assigned to Rodolfo Alonso (rodolfo-alonso-hernandez)
15:12:00 <ralonsoh> and the patch
15:12:03 <ralonsoh> #link https://review.opendev.org/#/c/690098/
15:12:15 <ralonsoh> this one is just failing randomly in the tests
15:12:30 <ralonsoh> but I didn't see any error related to the patch
15:13:02 <ralonsoh> (according to my understanding)
15:13:24 <ralonsoh> reviews are welcome always!
15:14:20 <slaweq> sure ralonsoh
15:14:24 <slaweq> I will review
15:15:25 <ralonsoh> ok, next seciton
15:15:32 <ralonsoh> #topic Open Discussion
15:15:40 <ralonsoh> I don't have anything here
15:15:43 <ralonsoh> and you slaweq ?
15:16:55 <davidsha> https://github.com/openstack/neutron/blob/master/neutron/plugins/ml2/drivers/openvswitch/agent/extension_drivers/qos_driver.py#L137
15:17:12 <ralonsoh> nice to see you
15:17:19 <ralonsoh> what is that?
15:17:22 <ralonsoh> ahhh yes
15:17:27 <davidsha> Hey, just got out of a meeting @.@
15:17:34 <davidsha> Sorry about that
15:17:52 <ralonsoh> the DSCP ofctl implementation
15:18:22 <davidsha> While I was working on the the classifier DSCP backend I realised that QoS and a few other extensions are not using the Native openFlow driver
15:18:38 <ralonsoh> instead of using the native implementation, the DSCP is using the cli one
15:19:00 <ralonsoh> I know this is another stone in the shoe
15:19:22 <ralonsoh> but we want to get rid of the CLI implementation and use only the native one
15:19:27 <davidsha> I'm working on a patch atm for the QoS to use the native driver, but we should probably try to deprecate: https://github.com/openstack/neutron/blob/master/neutron/agent/common/ovs_lib.py#L343
15:19:34 <ralonsoh> and you are going to refactor extensively the DSCP part
15:19:45 <ralonsoh> you should first use the native imp here
15:19:54 <davidsha> yup
15:20:32 <ralonsoh> I agree:  run_ofctl should be removed ASAP
15:20:52 <ralonsoh> I'll open a bug to track this, I'll send you the link
15:20:59 <davidsha> Will I include the deprecation flag in the patch?
15:21:04 <davidsha> Ack, thanks!
15:21:13 <ralonsoh> deprecation?
15:21:19 <ralonsoh> hmmmm
15:21:20 <davidsha> of run_ofctl
15:21:30 <ralonsoh> yes, we should
15:21:54 <ralonsoh> let me check how many calls in neutron are made to this function
15:22:37 <davidsha> will do, In Neutron, I don't think there are any after QoS, but I did see this being used in networking-sfc, I'll properly check Neutron.
15:23:11 <ralonsoh> this is used in Neutron just in OVSBridge
15:23:25 <slaweq> ouch
15:23:37 <slaweq> I didn't even know that we still have such method
15:23:38 <davidsha> yes, but the native is a child of OVSBridge
15:24:02 <ralonsoh> davidsha, yes, but only few methods using ofctl
15:24:09 <ralonsoh> the problem in sfc
15:24:11 <davidsha> yup
15:24:24 <ralonsoh> is sfc using OVSBridge methods?
15:24:41 <davidsha> SFC has one or 2 calls to it
15:24:44 <ralonsoh> or it's own implementation
15:24:53 <ralonsoh> to ovsbridge methods?
15:25:34 <davidsha> https://github.com/openstack/networking-sfc/blob/d30a2f576dc511295cbb29623dfa1b9bf84f5760/networking_sfc/services/sfc/common/ovs_ext_lib.py#L70
15:26:01 <davidsha> Line 57 above it proxies self to self.bridge
15:26:29 <ralonsoh> I understand they use it, it's easier
15:26:50 <ralonsoh> actually the way to run an os-ken application is not trivial
15:27:07 <davidsha> Ya, I'm still trying to get the DSCP mark to work
15:27:33 <ralonsoh> the good point in ovs dscp you are already inside an os-ken application
15:28:00 <ralonsoh> ok, let me open the bug to deprecate ofctl and I'll send you the link
15:28:27 <davidsha> kk, I need to drop, thanks!
15:28:39 <ralonsoh> bye
15:28:54 <davidsha> cya
15:28:56 <ralonsoh> anything else here in open discussion?
15:29:48 <ralonsoh> thank you all for attending and reading the log
15:29:58 <ralonsoh> see you here in two weeks
15:30:07 <ralonsoh> #endmeeting