15:00:40 <armax> #startmeeting neutron_drivers
15:00:41 <openstack> Meeting started Tue Nov 17 15:00:40 2015 UTC and is due to finish in 60 minutes.  The chair is armax. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:42 <mestery> amuller: lol :)
15:00:42 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:44 <openstack> The meeting name has been set to 'neutron_drivers'
15:00:57 <carl_baldwin> o/
15:00:59 <armax> mestery, amotoki, kevinbenton, carl_baldwin, HenryG ping?
15:01:07 <HenryG> yo
15:01:07 <amotoki> o/
15:01:08 * mestery waves at armax
15:02:00 <amuller> might as well stay and lurk :)
15:02:09 * ajo lurks
15:02:10 <armax> amuller: indeed
15:02:23 * regXboi lurks as well
15:02:43 <armax> https://bugs.launchpad.net/neutron/+bugs?field.status%3Alist=TRIAGED&field.tag=rfe
15:02:44 * regXboi asks amuller where the lurking corner is
15:02:57 <amuller> regXboi: on your left behind the water cool
15:03:09 <armax> we finished last week with bug #1513144
15:03:09 <openstack> bug 1513144 in neutron "Allow admin to mark agents down" [Wishlist,Triaged] https://launchpad.net/bugs/1513144 - Assigned to Carlos Goncalves (cgoncalves)
15:03:20 <armax> the submitter never came back
15:03:34 <armax> I am saying let’s mark this incomplete and move on
15:03:39 <amotoki> ah.. I forgot to comment more.
15:03:53 <amotoki> will follow up it after the meeting.
15:04:00 <mestery> armax: ++
15:04:21 <amotoki> marking this incomplete now sounds reasonable to me.
15:05:00 <carl_baldwin> +1
15:05:11 <armax> bug 1472704
15:05:11 <openstack> bug 1472704 in neutron "Support L3-only networking" [Wishlist,Triaged] https://launchpad.net/bugs/1472704 - Assigned to Neil Jerram (neil-jerram)
15:05:39 <armax> carl_baldwin: I am assuming the matching spec is the one you’re proposing, right?
15:05:52 <armax> or no?
15:06:06 <carl_baldwin> armax: That is Neil's but we're working to unify.
15:06:07 * regXboi goes to get a snack and some cold caffeine
15:06:43 <armax> carl_baldwin: so it should match this one, https://blueprints.launchpad.net/neutron/+spec/routed-networks
15:06:45 * carl_baldwin looks for the other rfe.
15:06:49 <garyk1> sorry i need to run
15:06:53 <mestery> bye garyk1
15:06:59 <armax> carl_baldwin: the other one was ‘segments’
15:07:02 <armax> carl_baldwin: I think
15:07:51 <carl_baldwin> armax: It all started here:  https://bugs.launchpad.net/neutron/+bug/1458890
15:07:51 <openstack> Launchpad bug 1458890 in neutron "Add segment support to Neutron" [Wishlist,Triaged]
15:08:18 <carl_baldwin> My spec is https://review.openstack.org/#/c/225384
15:08:42 <carl_baldwin> Trying to unify mine with the one from neiljerram_bb
15:08:43 <armax> right
15:09:08 <carl_baldwin> I will look for him today and hopefully get some updates to get them in line.
15:09:17 <armax> carl_baldwin: any progress on the one from neil?
15:09:35 <armax> carl_baldwin: I’d rather track everything with one bug report/spec if at all possible
15:09:41 <armax> carl_baldwin: unless the use cases are totally different
15:10:01 <armax> I’ll follow up on this one
15:10:07 <carl_baldwin> armax: ack  That is where we were headed.
15:10:15 <cgoncalves> Hi. on bug 1513144, apologies but I still need to catch-up on the comments yet. apologies
15:10:15 <openstack> bug 1513144 in neutron "Allow admin to mark agents down" [Wishlist,Incomplete] https://launchpad.net/bugs/1513144
15:10:30 <armax> cgoncalves: please do
15:10:37 <armax> and we’ll get back to you
15:10:53 <armax> bug 1512666
15:10:53 <openstack> bug 1512666 in neutron " Allow for per-subnet/network dhcp options" [Wishlist,Triaged] https://launchpad.net/bugs/1512666 - Assigned to Sam Betts (sambetts)
15:11:02 <cgoncalves> armax: thanks. it's been a crazy week last week :/
15:11:12 <armax> cgoncalves: for everybody, I can assure oyu
15:11:47 <armax> bug #1512666 seems trivial enough
15:12:35 <armax> carl_baldwin, ajo: do you happen to know if dnsmasq provides what we need?
15:12:55 <carl_baldwin> armax: I can't say without looking through dnsmasq docs.
15:12:56 * ajo reads, was in deep talk with obondarev about migrations
15:13:04 <amotoki_> one question is how we can distinguish a port which we need to apply the default dhcp options.
15:13:21 <ajo> armax , carl_baldwin , I will look at the dnsmasq config options too
15:13:26 <armax> amotoki_: I thought that port options take precedence
15:13:29 <ajo> I Can't recall something like that, but I don't know it all
15:13:33 <ajo> (about dnsmasq)
15:13:34 <amotoki_> extra_dhcp_opts is mainly used for ironic ports but not all ports are such ports.
15:14:05 <amotoki> but as RFE it is a reasonable request.
15:14:58 <armax> amotoki: agreed
15:15:07 <armax> amotoki: I wonder who could help to work on that
15:15:46 <amotoki> i can. I know ironic use cases and supported the inital extra dhcp opts work.
15:16:06 <amotoki> I am not hosting any blueprints.
15:16:44 <ajo> armax , I think dnsmasq provides what they're looking for, because they're already using dnsmasq
15:17:01 <armax> amotoki: cool
15:17:44 <ajo> it worries me how would we handle the allocation of new ips for unknown macs
15:18:00 <ajo> carl_baldwin , any thought on that?
15:18:22 <armax> ajo: agreed
15:18:31 <carl_baldwin> ajo It was my concern too but I haven't groked it enough to have further thoughts.
15:18:45 <ajo> neither me,
15:18:51 <armax> it sounds like we may want a spec for this, as there might be implicit assumptions that we’re overlooking here
15:19:04 <armax> we don’t want to start assigning dhcp addresses all over the place
15:19:19 <carl_baldwin> ++
15:19:24 <armax> ok
15:19:28 <amotoki> agree
15:19:31 <ajo> that could even mean, we may need to autodiscover new ports
15:20:09 <ajo> better to have a spec, I agree
15:20:29 <armax> ok, I’ll take care of it
15:20:35 <armax> next one
15:20:38 <armax> bug #1492714
15:20:38 <openstack> bug 1492714 in neutron "RFE: Pure Python driven Linux network configuration" [Wishlist,Triaged] https://launchpad.net/bugs/1492714 - Assigned to Li Ma (nick-ma-z)
15:20:51 <mestery> armax: I think that one is dependent on the privsep work
15:21:03 <mestery> And I know regXboi was looking to track that work which gus is doing
15:21:19 <regXboi> yes, this is dependent on privsep
15:21:25 <armax> right, it sounds like we all want this to happen
15:21:29 * regXboi looking where that is right now
15:21:31 <armax> but what it takes, I am not 100% sure
15:21:43 <HenryG> At the summit gus said he would soon be proposing patches to neutron
15:22:09 <HenryG> He said the oslo part is done (I think)
15:22:17 <mestery> HenryG: Wow, that's pretty awesome if it's done.
15:22:32 <salv-orl_> I don't think privsep is low-effort, but its development is probably mostly orthogonal. From what I gather it should not break anything...
15:22:37 <salv-orl_> ...until it's plugged in of course!
15:22:37 <regXboi> yes: https://review.openstack.org/#/c/238333/ merged last week
15:22:49 <mestery> salv-orl_: lol :)
15:23:12 <regXboi> current privsep patch is https://review.openstack.org/#/c/244984/
15:23:22 <ajo> I heard of a similar proposal which could be interesting
15:23:29 <ajo> I think it was apuimedon's
15:23:35 <ajo> moving the agents into the networking cgroup
15:23:41 <anteaya> well gus' work is also critical to nova, so if someone were to help his efforts I don't think he would say no
15:23:48 <ajo> so they have access to all networking related functions as-root
15:24:05 <mestery> ajo: With privsep, is that necessary still?
15:24:23 <ajo> mestery I guess it's an alternative to privsep
15:24:32 <ajo> or a separated thread can be moved into the cgroup
15:24:37 <ajo> it's an alternative
15:24:37 <mestery> ajo: Right
15:24:45 <regXboi> is it an alternative in oslo?
15:24:54 <regXboi> or is it project specific
15:25:16 <ajo> I think, it's something privsep is going to leverage :)
15:25:31 <ajo> sorry for the noise
15:25:49 <mestery> ajo: No noise, it's good to hear these other approaches too.
15:26:05 <mestery> ajo: Maybe it's worth commenting in the bug about this and adding apuimedon to it as well?
15:26:19 <salv-orl_> ajo: your punishment would be to contribute the work for assigning operations to cgroups rahter than run everything as root in oslo.privsep
15:26:20 <ajo> mestery : yes, doing
15:26:37 <mestery> sweet, thanks ajo
15:26:40 <ajo> salv-orl_ : lol :) I Will pass the punishment to apuimedon :D
15:26:44 <ajo> but I'd be glad to help
15:27:07 <salv-orl_> ajo: we're not in the army. if we were the army that would be the way to go
15:28:03 <regXboi> I see 1492714 as another step along the way to getting rid of shell
15:28:08 <armax> either way it sounds like this might be tricky to tackle in the mitaka timeframe
15:28:11 <regXboi> which is a Good Thing (TM) imho
15:28:27 <regXboi> armax: ++
15:30:45 <armax> ok, let’s move next, I’ll capture details on this conversation later
15:31:27 <armax> bug #1512864
15:31:27 <openstack> bug 1512864 in neutron "Application Metrics for Neutron" [Wishlist,Triaged] https://launchpad.net/bugs/1512864
15:31:39 <regXboi> armax: this needs to go to oslo
15:31:59 <mestery> regXboi: That was easy :)
15:32:02 <regXboi> :)
15:32:03 <armax> you mean bug 1512864?
15:32:07 <regXboi> armax: yes
15:32:21 <regXboi> that should be done across all of openstack, not just neutron
15:32:22 <mestery> regXboi: I suggest adding oslo to the affected projects then
15:32:25 <ajo> IMO, we better integrate with osprofiler first
15:32:37 <regXboi> mestery: agreed
15:32:49 <ajo> it does not cover continous profiling as they suggest, btw..
15:33:03 <armax> regXboi: ok, that makes sense
15:33:08 <ajo> yeah, and if we introduce profiling to openstack services, better to have something homogenous across services
15:33:09 <regXboi> ajo: yes that is something osprofiling needs
15:33:15 <ajo> that's a good point regXboi
15:33:28 <armax> and we could easily pull in stuff from oslo as it matures
15:33:43 <regXboi> armax: you want to change the bug or should I?
15:33:57 <salv-orl_> ajo: do we need to integrate with osprofiler? I usually just apply patches when I need to take measures and then collect data, I've honestly not yet seen a project with profiling in-tree
15:34:05 <armax> regXboi: change how? target oslo you mean?
15:34:18 <regXboi> armax: target oslo and s/neutron/openstack/ in the title
15:34:29 <ajo> salv-orl_ , I found it very interesting to debug RPC and flows between projects
15:34:30 <armax> regXboi: well
15:34:35 <armax> regXboi: the problem is:
15:34:43 <regXboi> salv-orl_: you *HAVE* to have profiling in tree if you are going to do ci/cd on performance
15:34:45 <salv-orl_> is "target oslo" needed because "all projects need it so there must be something for oslo too?"
15:34:46 <armax> regXboi: is the submitter even willing to work with oslo?
15:34:53 <ajo> salv-orl_ : good for profiling and reverse engineering for openstack newbiews (or outdated-oldbies..)
15:35:01 <armax> it may remain targeted and idle for ever
15:35:08 <regXboi> armax: the submitter is one of my co-workers - I can have a discussion with them to find out
15:35:34 <armax> regXboi: ok, let’s find out first also where the oslo project is at
15:35:34 <salv-orl_> regXboi: as long as you can give me a way to skip execution of profiling code that's fine.
15:35:46 <regXboi> salv-orl_: ++++++++
15:35:52 <ajo> salv-orl_ : +100
15:36:05 <armax> regXboi: it’d be tempted that we can simply say: this doesn’t belong to neutron and as such it’s ‘won’t fix’ in the current form
15:36:35 <armax> regXboi: let’s capture the details of this conversation and we’ll take it from there
15:37:12 <regXboi> armax: let me try to reach out today and have a conversation - there is a TZ differential and I'm driving up mestery's way in the afternoon/evening so that's best
15:37:36 <armax> regXboi: ok, no worries
15:38:05 <armax> does anyone fancy start going through the ‘confirmed’ list or shall we call it done, we’ll have a first pass offline and reconvene in a week?
15:38:23 <mestery> My vote: Offline and reconvene in a week
15:38:31 <HenryG> +1
15:38:43 <armax> ok, let’s make sure that we have a good triaged pipeline then
15:38:47 <regXboi> what about the new list?
15:38:48 <armax> I count on all of you
15:39:03 <armax> regXboi: we’ll have a pass too
15:39:09 <mestery> We won't let you down armax
15:39:39 <armax> mestery: I am sure
15:39:49 <carl_baldwin> ++
15:39:59 <mestery> armax: Thanks for waking up early to run these back-to-back meetings, we all appreciate yoru leadership!
15:40:04 <armax> ok, let’s go back at pounding on the keyboards
15:40:28 <armax> mestery: thanks for the encouragement but I am regretting this already :)
15:40:34 <mestery> lol
15:40:37 <amotoki> :-)
15:40:38 <armax> have a good tuesday everbody
15:40:39 <regXboi> armax: you can always move the meeting :)
15:40:47 <mestery> ++, thanks armax, you have a good day too!
15:40:52 <armax> or wednesday wherever you are
15:40:59 <armax> #endmeeting