16:01:21 <Sukhdev> #startmeeting ironic_neutron
16:01:24 <openstack> Meeting started Mon Jan  4 16:01:21 2016 UTC and is due to finish in 60 minutes.  The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:25 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:27 <openstack> The meeting name has been set to 'ironic_neutron'
16:01:43 <Sukhdev> Good morning folks -
16:01:56 <Sukhdev> who is here to attend the ironic-neutron meeting?
16:02:02 <Sukhdev> jroll : are you here?
16:02:08 <jroll> \o
16:02:21 <Sukhdev> lazy_prince, vsaienko ?
16:02:31 <vsaienko> hello to all
16:02:54 <Sukhdev> #topic: Agenda
16:03:01 <Sukhdev> #link: https://wiki.openstack.org/wiki/Meetings/Ironic-neutron#Meeting_January_4.2C_2016
16:03:21 <Sukhdev> #topic: Announcements
16:03:35 <Sukhdev> Happy New Year to everybody
16:03:53 <Sukhdev> hope everybody had a relaxing holidays!!
16:04:12 <vdrok> o/ happy new year!
16:04:43 <Sukhdev> I was trying to catch up on the action I missed in last two weeks -
16:05:17 <lazy_prince> 0/
16:05:33 <Sukhdev> lazy_prince : hi
16:05:47 <Sukhdev> #topic: Patches under review
16:05:50 <vsaienko> lazy_prince: hi
16:05:53 <Sukhdev> #link: https://etherpad.openstack.org/p/ironic-neutron-mid-cycle
16:06:15 * jroll wonders if devananda is around
16:06:49 <Sukhdev> I noticed there were bunch of updates to the patches - which is excellent
16:07:26 <lazy_prince> yup.. Thanks to vsaienko
16:07:35 <Sukhdev> and lots of review comments as well
16:07:51 <Sukhdev> yes noticed that vsaienko has been busy :-)
16:08:00 <Sukhdev> vsaienko : are we ready to ship :-)
16:08:10 <vsaienko> yep
16:08:27 <Sukhdev> vsaienko : perhaps you can give us an update and bring everybody to speed
16:08:57 <lazy_prince> that would be helpful assuming I missed couple of meetings..
16:09:13 <vsaienko> I've rebased  patches and resolved comments
16:09:42 <lazy_prince> is there anything left for me to work on.. :)
16:09:58 <vsaienko> I test all of them together, guide is on review also https://review.openstack.org/#/c/258596/7/doc/source/dev/ironic-neutron-integration.rst
16:10:33 <jroll> deva also mentioned he was able to test, mostly successfully
16:10:38 <vsaienko> I would be perfect if any body else also test them and provide feedback
16:10:57 <jroll> he found it wasn't isolating between two tenants, but it looks like the devstack stuff isn't intended to do so?
16:11:38 <vsaienko> jroll: it should
16:12:06 <lazy_prince> if tags are applied properly, they should be isolated..
16:12:07 <jroll> vsaienko: maybe it was because the default is a provider network
16:12:17 <vsaienko> from Newtron side each network in the tenant has unique VLAN
16:12:19 <jroll> I'd have to look back at irc logs
16:12:21 <Sukhdev> jroll : based upon the devstack patches, it should have
16:12:53 <vsaienko> ah, yes, I've missed part that configured network_provider in Ironic
16:13:15 <vsaienko> but already fixed it in last patchset, thanks vdrok!
16:13:40 <jroll> hrm, he said it isolated control plane okay
16:13:55 <jroll> anyway, not much use dwelling on that now
16:14:31 <Sukhdev> vsaienko : the list of patches on https://etherpad.openstack.org/p/ironic-neutron-mid-cycle is up-to-date, right?
16:15:16 <vsaienko> sukhdev: looks like yes
16:15:32 <Sukhdev> vsaienko : before I went on PTO, there was an issue with the devstack patches, which needed to be pushed again - wonder if those are listed correctly
16:15:48 <Sukhdev> I have not had time to verify recently
16:16:01 <devananda> o/
16:16:05 <jroll> they look right, Sukhdev
16:16:06 <vsaienko> hello devananda
16:16:10 <jroll> hai devananda :)
16:16:26 <devananda> Sukhdev, jroll: I tested the devstack support for that patch chain last week and confirmed it works
16:16:37 <devananda> nodes are moved between the provisioning provider network and the tenant network, etc, as expected
16:16:43 <jroll> \o/
16:16:44 <Sukhdev> devananda : Wow!! cool - thanks
16:16:51 <devananda> however I had difficulty creating >1 tenant network and separating traffic from tenant A and tenant B
16:17:01 <devananda> I suspect I was doing something wrong with regards to creating the neutron networks
16:17:43 <Sukhdev> devananda : oh I see -
16:17:46 <devananda> I started reviewing the patch chain over the weekend, but haven't gotten too far. at this point, it will be mostly nitpicks -- little things in the db code that I would like to see improved, etc
16:18:31 <Sukhdev> devananda : I wonder how you were creating neutron networks?
16:19:01 <devananda> as tenant A: neutron net-create; neutron subnet-create; nova boot ... --nic net=aaaa;
16:19:24 <devananda> so neutron assigned an IP from that range to the instance, but there wasn't actually an OVS network on the host that matched / routed to it
16:19:25 <Sukhdev> devananda : as an admin, when you create a network, you can specify which tenant it is for, etc..
16:19:47 <devananda> Sukhdev: does the admin need to create a network for each tenant (eg, when creating the tenant) ?
16:20:28 <vsaienko> devananda: but there wasn't actually an OVS network on the host that matched / routed to it what do you mean/
16:20:40 <devananda> vsaienko: yes
16:21:59 <Sukhdev> devananda : I believe so..
16:22:19 <devananda> first off, I created two tenants, did not define any networks for them, and just booted two servers. Neutron assigned a private IP to each one -- on the same subnet
16:22:24 <devananda> and I was able to route traffic between them
16:22:44 <devananda> so while provisioning network was not visible to tenants, they were visible to each other
16:23:21 <devananda> then, as each tenant (not the admin), I issued "neutron net-create ...; neutron subnet-create ...;" commands with non-overlapping IP ranges
16:23:39 <devananda> and booted a server on that network
16:24:02 <devananda> neutron assigned an IP from the new subnet range, but it was inaccessible.
16:24:22 <vsaienko> devananda: did you checked that VLAN ID has been changed in OVS with ovs-vsctl show?
16:24:38 <devananda> vsaienko: I did not see any new networks on the host
16:25:36 <Sukhdev> devananda : your sequence of steps seems right
16:26:30 <vsaienko> devananda: that is trange, since once you create new neutron network, neutron should create new IP namespace for it
16:26:34 <Sukhdev> only difference is that I would create a networks from admin context and specify the tenant for these networks and then create subnets from the tenant's context
16:26:37 <vsaienko> *strange
16:27:00 <Sukhdev> but, your sequence should work as well
16:27:32 <devananda> Sukhdev: if this were hardware switches instead of OVS, in a real deployment, would the non-admin tenant be able to create a neutron network?
16:27:57 <Sukhdev> devananda : yes
16:27:59 <vsaienko> devananda: it should
16:28:04 <devananda> hmm ...
16:28:21 <vsaienko> neutron should call create_network() from ML2 driver and it will do all the rest
16:28:47 <vsaienko> I mean ML2 should create missing VLANs on the switches
16:28:53 <Sukhdev> devananda : I think you are doing everything correct
16:29:17 <devananda> I see
16:29:35 <devananda> that's good - but then I don't know why it didn't work locally
16:29:35 <Sukhdev> devananda : we need to triage a bit to dig deep - as vsaienko pointed, we need to check if correct vlan ID are being used
16:30:05 <devananda> I don't know if I'll have time to rebuild that env today, but I'll try to help as much as I can
16:30:29 <Sukhdev> devananda : are you using HW switches or OVS ?
16:30:32 <devananda> OVS
16:30:39 <devananda> I don't have that kind of fancy hardware at home ;)
16:31:21 <Sukhdev> devananda : in that case, I think OVS plumbing may be missing something -
16:31:27 <Sukhdev> ovs-vsctl show output will be helpful
16:31:47 <vsaienko> devananda: if you still have environment, I'm ready to debug issues with you
16:32:20 <devananda> vsaienko: I do not
16:33:07 <Sukhdev> I will set up the test setup this week and play with it as well - and will ping you guys with the feedback
16:33:38 <Sukhdev> vsaienko : did you test the same way as devananda is trying to do?
16:33:45 <vsaienko> yes
16:33:50 <Sukhdev> hmmm...
16:34:24 <Sukhdev> OK - lets try to reproduce it then....I will get on it this week as well
16:34:47 <devananda> Sukhdev: what I want to get as output of this test is confirmation of isolation between tenants
16:34:50 <vsaienko> I will triple check all
16:35:29 <Sukhdev> devananda : understood - and, actually, that leads me to the next topic on agenda
16:35:56 <Sukhdev> Is Yuri here?
16:36:15 <vsaienko> sukhdev: he is on PTO this week
16:36:21 <Sukhdev> he was writing a tempest scenario test to test exactly the same thing
16:36:31 <devananda> \o/
16:36:41 <vsaienko> https://review.openstack.org/#/c/258934/
16:36:47 <Sukhdev> devananda : see here - https://bugs.launchpad.net/tempest/+bug/1520230
16:36:48 <openstack> Launchpad bug 1520230 in tempest "Test Case: Create a test scenario to verify that Ironic supports multitenancy" [Undecided,In progress] - Assigned to Yuriy Yekovenko (yyekovenko)
16:37:49 <Sukhdev> vsaienko : do you know if he was able to test this patch?
16:38:16 <devananda> jroll: aren't we moving tempest scenario tests into ironic's tree soon?
16:38:21 <Sukhdev> #link: https://review.openstack.org/#/c/258934/
16:38:34 <vsaienko> as far I know he is still working on that patch
16:38:43 <jroll> devananda: that's a goal, yes
16:39:38 <devananda> yea, found it
16:39:47 <devananda> vsaienko: he may want to rebase on top of https://review.openstack.org/#/c/253982/ or something
16:40:27 <vsaienko> devananda, I will ping him about it
16:41:07 <jroll> devananda: I'm curious what other work is needed there, but probably a good idea
16:41:11 <Sukhdev> Looks like this test will be very helpful in testing this scenario
16:41:21 <devananda> jroll: I think we were bike shedding on the folder name
16:41:38 <jroll> devananda: oh good
16:41:49 <jroll> devananda: I meant in project-config and such though :)
16:41:50 <devananda> yep. https://review.openstack.org/#/c/246161/7
16:42:00 <devananda> jroll: oh. that I'm not sure about
16:42:10 <jroll> devananda: yeah, it's fine though
16:43:35 <devananda> i'm not saying we shouldn't land the multitenancy testing in tempest - just making sure folks are aware of the move to tempest-lib
16:45:50 <Sukhdev> I am hoping Yuriy is aware of this tempest-lib stuff
16:46:24 <Sukhdev> it has been a long time since I touched tempest stuff - I am not up-to-date in that regard :-):-0
16:46:58 <Sukhdev> For now, as long as we can test multi-tenancy manually, we should be good
16:47:21 <Sukhdev> Anything else to discuss on this topic?
16:47:55 <Sukhdev> #topic: Integration Status
16:48:27 <Sukhdev> We have mostly covered the integration status already, but, I wanted to specifically talk about item 7 on https://etherpad.openstack.org/p/ironic-neutron-mid-cycle
16:49:00 <Sukhdev> This is based upon two week old information - perhaps this has been addressed while I was gone
16:49:05 <Sukhdev> any update on this?
16:49:29 <jroll> that patch still needs a major update
16:49:36 <jroll> it's on my todo list, apologies for not getting to it
16:50:47 <Sukhdev> jroll: this will impact the portgroups, which no body is testing as of now
16:50:49 <lazy_prince> i am sorry.. which item 7.. under issues o rpatches..?
16:50:57 <Sukhdev> lazy_prince : yes
16:51:08 <jroll> Sukhdev: yes I know
16:51:09 <Sukhdev> lazy_prince : issues
16:51:15 <lazy_prince> aha.. okay..
16:51:51 <vsaienko> folks, at the moment we don't test portgroups at all, OVS allows to create portchannels with LACP I think we should try to cover portgroups with OVS also
16:51:59 <Sukhdev> OK - I was just making sure that I am up-to-date on this status
16:52:34 <lazy_prince> vsaienko: ironic code is not upto date with portgroups yet.. we will have to wait for that...
16:53:47 <Sukhdev> The question for the team is - should we consider merging the patches for M2 (mid Jan)?
16:54:05 <Sukhdev> and do the portgroups after that for M3?
16:54:18 <lazy_prince> where are we with nova patches...?
16:54:20 <jroll> we should merge all the things as soon as they're ready :P
16:54:38 <jroll> lazy_prince: 2 patches, one has a +2, the other is the portgroups thing that needs a major update
16:54:42 <lazy_prince> before we merge, we need to be ready from nova side too..
16:55:00 <lazy_prince> cool..
16:55:29 <Sukhdev> update on etherpad is two weeks old - but, one of the patches was merged back then
16:55:43 <vsaienko> I'm asking to review devstack related patches so we have consensus on global variables names and can proceed with https://review.openstack.org/#/c/252814/
16:56:20 <jroll> vsaienko: will do
16:56:30 <vsaienko> jroll: thanks~
16:56:52 * Sukhdev time check - 4 min left
16:57:21 <Sukhdev> lazy_prince : have you had a chance to test the devstack patches?
16:57:47 <devananda> Sukhdev: I only saw one devstack patch when I was testing last week?
16:58:19 <lazy_prince> hmm... was busy all this time but free from now on.. I will test and update...
16:58:21 <Sukhdev> devananda : I try to keep the list of all patches on - https://etherpad.openstack.org/p/ironic-neutron-mid-cycle
16:58:59 <Sukhdev> hence, I was asking earlier in the meeting to make sure the team updates this etherpad -
16:59:08 <Sukhdev> this keeps things simple to track
16:59:16 <devananda> ah, thx
16:59:17 <jroll> devananda: there's a bunch in the chain
17:00:10 <Sukhdev> Lets all try to review devstack patches this week
17:00:21 <Sukhdev> We are out of time - folks
17:00:29 <Sukhdev> thanks for attending - this was great discusion
17:00:30 <krotscheck> o/
17:00:34 <Sukhdev> bye
17:00:39 <Sukhdev> #endmeeting