16:01:34 #startmeeting fuel 16:01:34 #chair xarses 16:01:34 Todays Agenda: 16:01:34 #link https://etherpad.openstack.org/p/fuel-weekly-meeting-agenda 16:01:34 Who's here? 16:01:35 Meeting started Thu Feb 25 16:01:34 2016 UTC and is due to finish in 60 minutes. The chair is xarses. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:01:36 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:01:38 The meeting name has been set to 'fuel' 16:01:40 Current chairs: xarses 16:01:43 hi 16:01:44 hi 16:01:45 hi 16:01:46 hi 16:01:48 hi 16:01:48 hi 16:01:50 hi 16:01:54 hi 16:01:56 hi 16:01:57 hi 16:02:02 hi 16:02:03 hi 16:02:14 hi 16:02:16 hi 16:02:16 o/ 16:02:19 #topic action items from last week 16:02:28 ashtokolov will look into increased deployment time in smoke tests 16:02:29 \o/\o/ 16:02:37 o\ 16:02:41 hi 16:03:05 o/ 16:03:21 o/ 16:03:29 ashtokolov: is typing 16:03:32 hi 16:03:33 please be patient 16:03:33 =) 16:03:39 .... 16:03:41 we discovered one issue with apache task https://review.openstack.org/282455 16:03:53 hi 16:04:01 hi there 16:04:42 it was a floating issue, fix was merged 16:04:57 so did deployment time decrease? 16:05:03 or are we still facing issues there? 16:05:42 .... 16:05:49 ashotokolov is typinh 16:05:51 .... 16:06:13 .. by letf foot :P 16:06:22 xarses, deployment time decreased by 5-10% 16:06:37 as we expected with 4-core cpus of CI nodes 16:06:43 sorry folks, slow wi-fi in the office 16:06:58 slower than you typing? :P 16:07:06 .... 16:07:12 ashtokolov is typing 16:07:14 .... 16:07:18 ok, moving 16:07:24 ikalnitsky holser to follow up with dnikishov on the spec for non-root https://review.openstack.org/#/c/243340/ 16:07:49 we have reviewed the spec 16:07:58 i think the main approach is established 16:08:07 so i start merging patches to fuel-web 16:08:29 thanks 16:08:41 fzhadaev yottatsa to sync on NFV status and provide combined update 16:08:49 Done. And vsakharov will provide our combined update 16:08:55 :) 16:09:24 thanks 16:09:25 it's in agenda 16:09:28 #topic code review backlog status: http://bit.ly/1Kp8BzM (mihgen) 16:09:59 folks we lag behind 16:10:14 I won't be giving actions items, but please check the link and help to move this forward 16:11:14 Yes, we want to keep up on our code review throughput 16:11:30 thanks mihgen 16:11:46 #topic Using multiple OpenStack versions https://review.openstack.org/#/c/281557/ (xarses) 16:12:03 oh, we've scoped it down a bit since I wrote that =) 16:12:07 I've been working to socialize https://review.openstack.org/#/c/281557/ so we can start versioning openstack tasks seperate of others. This led to needing to open https://review.openstack.org/#/c/284294/ (remove some of the ::openstack manifests) this also relates to the prior because some of the calls are inderected in ::openstack before making it to the puppet-openstack modules. 16:13:15 questions? 16:14:09 #topic feasibility of converging with upstream https://launchpad.net/openstack-resource-agents (aspiers) 16:14:15 hi 16:14:24 thanks for tolerating an outsider crash your meeting :) 16:14:34 I am the maintainer of this upstream projecct 16:14:36 welcome 16:14:46 and I recently noticed that fuel has forked some of the OCF RAs 16:14:59 holy guacamole, Adam! welcome! 16:15:01 I talked to bogdando and he suggested I raise the idea of reconvergence here 16:15:04 :) 16:15:17 hi 16:15:22 I think it could be a win-win situation for all 16:15:28 by combining efforts 16:15:31 I'm for it, we have a bunch of stuff in here 16:15:38 that is diverged 16:15:50 I am guessing the fork happened whilst the project was unmaintained and living in github.com/madkiss 16:15:51 and bunch of improvements :) 16:16:07 yes, few years ago 16:16:11 but since then I brought it into the openstack ecosystem and we have various vendors contributing now 16:16:11 forgive me now knowing, what do you have for CI coverage? 16:16:32 we already have the OCF installed as a RPM, so we could swap them 16:16:37 xarses: nothing right now but we have just started preliminary work on fixing that 16:16:49 e.g. https://launchpad.net/bugs/1508559 16:16:49 Launchpad bug 1508559 in openstack-resource-agents "Needs bashate CI" [Wishlist,Confirmed] 16:16:49 aglarendil is typing 16:16:56 evntually, we'd want to vote, at least in regards to what we consume 16:17:26 aspiers: have you seen https://ci.fuel-infra.org/view/puppet-openstack/ ? 16:17:26 we have a bunch of CI around beating the cluster 16:17:32 oh, common, is there anyone using non-bash shell? 16:17:35 note, I have a PoC Travis CI for docker containers running a pacemaker cluster 16:17:43 angdraug: no I haven't 16:17:43 aglarendil: only all of Debian users :) 16:17:46 this may be used as the base for OCF under tests 16:17:49 bogdando: cool! 16:17:51 or suchlike 16:17:55 I think we definitely want unit tests too 16:18:11 angdraug: the first thing you do 'dpkg-reconfigure dash' 16:18:11 although that would require mocking dependencies, and it's a pain with shell code 16:18:29 but anyway I don't mean to hijack your meeting with our future plans 16:18:41 example OCF testing job https://travis-ci.org/bogdando/rabbitmq-server/builds/109353708 16:18:46 just wanted to raise it and see if convergence potentially appeals to you guys 16:18:47 JFYI 16:19:06 aspiers: lets prepare a ML or such, and find out who can work on this, and then we should maybe setup a meeting time to talk just about this 16:19:08 my vote is +1. Although we need specs submitted, both sides 16:19:08 since I think these RAs definitely have potential to be shared across many vendors 16:19:10 looks like everyone is excited about the prospect of reconvergence 16:19:41 xarses: makes sense 16:19:46 aspiers: can I put an action on you to follow up on this? 16:19:47 bogdando: would you be our project's liaison with openstack-resource-agents? 16:19:51 xarses: sure 16:19:52 I am +1 for convergence 16:20:07 angdraug, ok 16:20:28 #action aspiers will create ML to find out who is interested in RA convergance and probably set up dedicated meeting for such 16:20:33 ok moving? 16:20:37 yup 16:20:46 thanks! 16:20:57 thanks for raising this aspiers 16:21:01 #topic topics for design summit in Austin https://etherpad.openstack.org/p/fuel-austin-agenda (angdraug) 16:21:12 #link https://etherpad.openstack.org/p/fuel-austin-agenda 16:21:12 me and mihgen has added session leads and required participants to the proposed topics for Austin 16:21:12 please review and if you find your name on the list confirm that you can lead discussion of the topic 16:21:12 if you add another topic, please leave your name in brackets so that we don't have to guess who added it 16:21:15 if you see topics that are better served by online discussion or spec review, feel free to flag them 16:21:18 we will finalize the high-level agenda next week, so don't wait until the week before the summit to review this 16:21:21 thanks 16:22:20 moving? 16:22:41 angdraug: do we have right to propose some changes? 16:22:46 of course 16:22:52 how to > confirm that you can lead discussion of the topic ? 16:23:08 vkramskikh: I think silent consent is enough 16:23:21 object if you see your name against something you don't know enough about 16:23:37 or if you think someone else should be there 16:24:23 aglarendil: what kind of changes do you have in mind? 16:24:46 angdraug: finalizing HA reference architecture with power management and proper event-based control 16:25:23 aglarendil: add a ~victim~ owner 16:25:31 me! 16:25:37 mi-mi-mi 16:26:10 ok moving 16:26:17 #topic Plugins v5 status (ikutukov) 16:26:29 Plugins v5 features: 16:26:40 - deployment tasks v2.0.0 is used as only task version 16:26:44 ikalnitsky asked us to implement: 16:26:47 - per-release configuration including deployment graph 16:26:50 And request from our services: 16:26:54 - configuration folders and custom paths support 16:26:57 Spec review: 16:27:00 https://review.openstack.org/#/c/271417/ 16:27:48 This came up form a conversation that I had with some one working on a plugin. The wanted to know when the v5 version and tags would be ready so that they can finalize the metadata for their 9.0 plugins 16:28:12 will we be able to make it before FF.. ? 16:28:14 xarses: tags? you meant repo tag? 16:28:38 well as of yesterday, 5.0 wasn't merged 16:28:44 mihgen: i think only nailgun part must be ready before ff. fuel-plugin-builder has untied release cycle 16:28:55 in fuel-plugins or rpm for fuel-plugin-builder 16:29:00 xarses: yeah. i want to release current master on pypi 16:29:18 i'm waiting for merging my patch to openstack-infra 16:29:30 that pushes new rlease to pypi once tag is pushed 16:29:40 ikalnitsky: also, I noticed we don't have recent tags or branches for 7.0 or 8.0 16:30:02 xarses: that's because we have untied from Fuel release cycle 16:30:31 but there are no tags for 4.0 then 16:30:38 yes 16:30:44 #ikalnitskiy its hard, but there are some chances to make nailgun support before SCF, some part of the FPB is on review but it's not possible to finish work on FPB before SCF 16:30:50 i will push it once this patch is merged 16:30:56 #link https://review.openstack.org/#/c/283683/ 16:31:48 wait, 4.0 version on fuel-plugins == v5 plugin support? 16:31:58 nope 16:32:06 4.0 = v4 plugins 16:32:13 make a release on pypi 16:32:18 merge v5 to master 16:33:07 ok, this is all confusing 16:33:33 I will follow up after the meeting 16:33:39 ok 16:33:44 #topic move ceph/public to Fuel Storage by default https://review.openstack.org/#/c/284467/ (agrebennikov) 16:33:56 It happened on the number of customers where we had to do dirty hack and replace ceph/public and ceph/replication net assignments to fuel networks 16:34:03 огые нуыеуквфн ыфьу куйгуые 16:34:07 ooops)) 16:34:16 just yesterday - same request one more time 16:34:24 is possible to do it with templates, but there is no reason to ever mix item with openstack management 16:34:32 usually customers have slow interfaces for public/mgmt and fast interfaces for storage. so lets place entire ceph traffic to storage and let the user to split ceph traffic with net templates when necessary 16:34:52 objections anyone? 16:35:08 +1 from me 16:35:18 basically ceph doesn't even Require cluster network 16:35:43 do we have a bug for it.. ? 16:35:45 but xarses said it is impossible to test cases "have cluster"/"no cluster net" 16:35:50 we do 16:35:52 mihgen: linked from the commit 16:36:02 I'd get ceph folks to review.. 16:36:22 ok 16:36:26 mihgen: please invite the right people to +1 16:36:32 I'll follow up with them 16:36:56 thanks for raising this agrebennikov 16:37:04 np 16:37:16 thanks guys, leaving 16:37:20 #topic UI Team status (vkramskikh) 16:37:26 Hi! Here is our status for 9.0 features: 16:37:26 1) https://blueprints.launchpad.net/fuel/+spec/remove-vendor-code - the removal request has been merged; request to restore them in the downstream is still in progress - expected to be merged soon after restructuring the code. 16:37:26 2) https://blueprints.launchpad.net/fuel/+spec/allow-choosing-nodes-for-provisioning-and-deployment - the first part (separate provisioning) was merged, separate deployment and choosing nodes are going to land soon. 16:37:26 3) https://blueprints.launchpad.net/fuel/+spec/separate-fuel-ui-repo - separate test runner was merged, but still doesn't work in case of separate repo; still unsure if we could make it before FF. 16:37:29 4) NFV stuff - implementation for node attributes (for Nova and DPDK CPU pinning) is close to be implemented and merged; other changes (topology representation and interface screen changes) still require working API before we can merge UI changes. 16:37:35 Questions? 16:39:06 seems no, thanks for the update 16:39:07 >>> but still doesn't work in case of separate repo 16:39:16 i 'll check it later today 16:39:36 i believe nothing serious 16:39:56 thanks kozhukalov 16:40:00 moving on? 16:40:13 #topic Bugfix team status (dpyzhov) 16:40:19 hi guys 16:40:31 we all good with 8.0 release, as you know 16:40:58 for high priority bugs we have constant number of bugs in python 16:41:09 angdraug: releasing?.. 16:41:43 and we solving library bugs as they appear 16:41:58 but we have lack of python guys right now 16:42:05 and we have 100 medium priority bugs in python 16:42:13 and 3 weeks before SCF 16:42:28 :( 16:42:51 library guys are helping with python 16:43:09 but they have limited expertise and also they have their tasks 16:43:22 mihgen: yes on releasing 8.0 16:43:27 bookwar: ^ 16:43:43 it doesn't look like we'll be able to get good progress with medium priority bugs in 9.0 and most of them will be moved to 10.0 16:44:32 we don't have much time to introduce new medium bugs, either... 16:44:54 dpyzhov: it sounds like we need to prioritize them over an iteration in 10, since we will have a longer cycle 16:44:55 2/3 of medium priority bugs were moved from 8.0 release 16:45:43 dpyzhov: do you mean 1/3 of medium bugs were fixed in 8.0, or that 1/3 of currently open medium bugs were introduced in 9.0? 16:45:55 folks there many other topics in agenda, time.. 16:46:06 lets try to work out how we can knock them down in 10, we don't have time for 9.0 either 16:46:07 1/3 of currently open medium bugs were submitted after 8.0 SCF 16:46:27 that's all that I wanted to highlight 16:46:55 #topic Fuel + Solar integration (pigmej) 16:47:06 Hey 16:47:11 As you may now we should integrate solar with fuel in 9.0. Integration in 9.0 will be very experimental and it will not affect fuel except: https://review.openstack.org/283600. 16:47:11 We will introduce 2 new packages (and some dependencies) in Centos7, these dependencies are already packaged and available in mos-master. 16:47:11 Predicted UX: 16:47:11 - user will need to install 2 rpm packages: solar and fuel2solar 16:47:13 - user will configure everything in fuel-web and then switch to solar CLI before clicking "deploy" button 16:47:16 16:47:19 Current plan is to: 16:47:22 1. create fuel-spec about introducing these 2 additional packages (and requirements), this spec will not requre any QA activities, 16:47:25 2. then should we describe everything again in openstack-dev ML 16:47:28 3. describe this everything in fuel docs 16:47:31 16:47:35 16:47:37 Or maybe instead fuel-spec in 1 could we just create blueprint with pretty much the same content ? 16:49:11 pigmej do we need to land any patches to nailgun before FF to make it work? 16:49:15 do we even need to introduce the packages onto the ISO? can we just leave them in the fuel mirror? 16:49:28 another q is should we put these two packages on the iso (i think we should) 16:49:32 ashtokolov: https://review.openstack.org/283600 this is only one required 16:49:33 if it's just about adding a couple of packages to the mirror, then I don't think we need a spec for it 16:49:34 ashtokolov: one patch, that extends nailgun api 16:49:49 ashtokolov: i believe same thing you was describing 2 months ago 16:49:58 solar package is already in mos-master, requirements are there also 16:50:14 xarses: by default we use local mirror (unfortunately) 16:50:33 so we'd better place these packages on iso 16:50:44 kozhukalov: we have fuel centos mirror on the fuel-master 16:50:51 dshulyak you mean serialised graph and deployment info? 16:51:12 even with deb, we still have the fuel-mirror in the repo list 16:51:38 for now we have just rpms 16:51:38 anyway, we need to move 16:51:53 thank pigmej 16:51:57 #topic Telco Team Status + NFV (vsakharov) 16:52:02 Hi all. 16:52:03 For now, Telco team has three main activities for 9.0: 16:52:03 1) Support NFV features: 16:52:03 1.1) Huge pages [1] 16:52:03 1.2) NUMA/CPU pinning [2] 16:52:03 xarses: you mean let's leave this to a user to download these packages using fuel-mirror? 16:52:03 1.3) SR-IOV [3] 16:52:05 ashtokolov: yes, but deployent info is already available using existing handlers, that patch adds is only for serialized graph 16:52:05 1.4) DPDK [4] 16:52:07 All specs were merged. 10 patches merged, 17 - on review 16:52:09 2) Daemon Resource Allocation Control [5] 16:52:09 sorry 16:52:11 Spec is merged. 4 patches are on review. 16:52:13 3) Removing Mirantis-specific code from fuel code 16:52:15 3 patches are merged. 16:52:17 [1] https://blueprints.launchpad.net/fuel/+spec/support-hugepages 16:52:19 [2] https://blueprints.launchpad.net/fuel/+spec/support-numa-cpu-pinning 16:52:21 [3] https://blueprints.launchpad.net/fuel/+spec/support-sriov 16:52:23 [4] https://blueprints.launchpad.net/fuel/+spec/support-dpdk 16:52:25 [5] https://blueprints.launchpad.net/fuel/+spec/cgroups 16:52:27 Any questions? 16:52:57 kozhukalov: yes, we don't need to include it in the release that way, its just packages sitting on the internet 16:53:20 pigmej: I think you'd at least need to raise this in ML, and then see if spec/whatever is needed.. 16:53:29 vsakharov: whats at risk for feature freeze? 16:54:00 DPDK with high risk 16:54:00 mihgen: I will post the same on ML then, and we will see what will happen. 16:54:13 all other fuel related - with medium 16:54:28 merging is going a bit slow :( 16:54:36 xarses we're expecting to merge library and agent #link http://bit.ly/1oCR2Jt before FF 16:54:45 #action pigmej will post on the ML with regards to solar packages merge 16:54:56 any creative ideas to metigate it / speed up review/merge? 16:55:09 vsakharov: dkalleg yottatsa thanks 16:55:57 #topic Network team status (alex_didenko) 16:56:10 Our team is working on the following features. 16:56:10 https://blueprints.launchpad.net/fuel/+spec/allow-any-vip - requires some changes in spec and design, working on them, patches are on review. Spec change and email to ML are in TODO. This feature looks good so far. 16:56:10 https://blueprints.launchpad.net/fuel/+spec/support-sriov - Only library patches are in good shape, no patches merged in other projects. So this feature is in red state (risk of not landing till FF is above average) 16:56:10 https://blueprints.launchpad.net/fuel/+spec/support-dpdk - Even library patches are not yet ready to be merged (but they are on review though). This one is red also (risk of not landing till FF is very hight). 16:56:22 some more details on NFV :) 16:57:03 Done 16:57:15 yep, we're stuck with nailgun 16:57:42 yottatsa: what's that? 16:58:26 there are 5 topics left. Should those be followed up over email, since we won't be able to discuss all of them now.. ? 16:58:27 #topic Nonroot feature status (dnikishov 16:58:34 hi 16:58:43 _ hi 16:59:01 as ikalnitsky mentioned, there are few merged patches for fuel-web 16:59:08 mihgen: probably 16:59:30 the plan is to get non-vital commits merged first and coordinate the rest with infra 16:59:57 are you getting help for library patches? 17:00:19 we are out of time guys 17:00:19 yes, I've got some reviews 17:00:36 please follow up on the remaining items on the ML 17:00:39 here's the list of patches to be merged in first order and in second 17:00:41 #link https://etherpad.openstack.org/p/nonroot-safe-merging 17:00:43 o/ 17:00:49 hi * 17:00:56 #endmeeting