16:03:46 <vkozhukalov> #startmeeting Fuel 16:03:47 <openstack> Meeting started Thu Apr 3 16:03:46 2014 UTC and is due to finish in 60 minutes. The chair is vkozhukalov. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:03:48 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:03:50 <openstack> The meeting name has been set to 'fuel' 16:04:00 <vkozhukalov> Greeting everyone 16:04:04 <meow-nofer> sup 16:04:07 <vkozhukalov> who is here? 16:04:11 <agordeev> o/ 16:04:13 <mattymo> mattymo is here 16:04:32 <aglarendil> me 16:04:41 <evgeniyl> I'm here 16:05:14 <vk> me 16:05:20 <meow-nofer> let's discuss something 16:05:31 <xarses> here 16:05:57 <vkozhukalov> #topic announcements 16:06:27 <vkozhukalov> Do you guys have any announcements? 16:07:03 <meow-nofer> we actually had a meeting on how we're gonna implement UI plugins for Nailgun 16:07:39 <vkozhukalov> have any results been achived? 16:08:00 <meow-nofer> yep, I hope Vitaly will provide some links to docs 16:08:05 <vk> yep, today we discussed approach described here: https://docs.google.com/a/mirantis.com/document/d/18UwHEWO2AwHf84HTS-wDmiceNo68wUg7r24xcI_Jhec/edit# and all of us found this approach viable 16:09:02 <vkozhukalov> Ok, looks like we have design for UI plugins 16:09:05 <meow-nofer> I'm going to write some similar documentation on Nailgun API/Backend plugins, as well, but not for pending release, obviously 16:09:05 <vkozhukalov> Great 16:09:12 <vk> yep, nailgun part to go 16:09:37 <meow-nofer> there are no known unsolved issues right now as far as I know 16:09:46 <vkozhukalov> any other announcements? 16:10:00 <xarses> we have merged single controller ha, I have started working on updating the fuel-main tests and hope to have something ready by tomorrow 16:10:45 <meow-nofer> also, Pshemek introduced some great possible improvements in how we work with networks 16:10:54 <vkozhukalov> xarses: wow, we got rid of simple mode 16:10:57 <vkozhukalov> right? 16:11:07 <meow-nofer> this all is described in our meeting notes and hopefully soon be in fuel-dev 16:11:12 <dpyzhov> preparation to 4.1.1 release is in progress. we have only two unmerged requests in web part: https://launchpad.net/fuel/+milestone/4.1.1 16:11:18 <xarses> vkozhukalov: it's still there for now, but ha_compact is the default 16:11:55 <vk> xarses, should we now remove any mentions of deployment mode from UI? 16:11:59 <vkozhukalov> meow-nofer: will you send a letter about UI meeting? 16:12:51 <meow-nofer> I guess I'll pass this honor to Vitaly 16:13:04 <xarses> vk, We will, but we need to give some time to using ha_compact by default to ensure it's stable 16:13:06 <meow-nofer> let's keep all our discussions on this in fuel-dev 16:13:36 <vk> I'll also try to send description of networking-related stuff we decided to implement by tomorrow 16:14:01 <meow-nofer> I mean both fuel-dev channel and mailing list 16:14:11 <vkozhukalov> ok, let's move on 16:14:38 <vkozhukalov> #topic Blueprints 16:15:20 <vkozhukalov> #link https://blueprints.launchpad.net/fuel/+spec/fuel-containerization-of-services 16:15:46 <vkozhukalov> what is the status? 16:16:22 <aglarendil> we have barely 90% complete 16:16:25 <mattymo> vkozhukalov, you want to hear from me 16:16:53 <aglarendil> I would rather ask Matthew mattymo to provide an update 16:16:53 <vkozhukalov> mattymo: it would be great to hear from anyone ) 16:16:59 <mattymo> We have prototypes of nearly every service ready, but a few aren't puppetized. A few of us are trying to solve a bunch of tasks in parallel 16:17:08 <mihgen> aglarendil: what about cobbler dnsmasq? 16:17:16 <mattymo> aglarendil is working on forwarding DHCP broadcast packets via iptables mirroring rules 16:17:25 <mattymo> but it looks like we need a 3.10 kernel for that feature 16:17:35 <aglarendil> I think this will be our best shot 16:17:49 <mattymo> I'm working on Fuel ISO build tools to prepare an ISO that has all the container images bundled (plus a lightweight tool to deploy them after) 16:17:52 <mihgen> aglarendil: can we implement it without requirement of 3.10? 16:17:56 <xarses> whats wrong with dnsmasq? 16:18:05 <mattymo> xarses, dnsmasq is beautiful 16:18:12 <aglarendil> we need dnsmasq to receieve broadcasts in docker container 16:18:32 <mattymo> docker sets up a NAT network inside of an ip network namespace. it can't hear the DHCP broadcast requests from nodes on the physical LAN 16:18:49 <xarses> dhcp-relay? can dhcp-helper in the parent hep? 16:18:58 <xarses> dhcp-relay? can dhcp-helper in the parent help 16:19:38 <aglarendil> it is another option 16:19:45 <xarses> actually sounds exactly like what i've been using dhcp-helper to do 16:19:54 <mihgen> this would not require 3.10, right? 16:20:00 <mattymo> Alexander Didenko is working on puppetizing some of the remaining services (rsyslog, postgres, rabbitmq) 16:20:13 <xarses> correct 16:20:57 <xarses> example i've been using for multi-l3: dhcp-helper -s 10.110.0.2 -i virbr6 -i virbr11 16:21:27 <vkozhukalov> are there any questions about this blueprint? moving on? 16:22:10 <vkozhukalov> #link https://blueprints.launchpad.net/fuel/+spec/patch-openstack 16:22:49 <aglarendil> we have puppet providers complete. now we are working on the Astute layer that does repository update on the nodes 16:23:03 <vkozhukalov> looks like 50% completed. right? 16:23:33 <mihgen> what about multiple repos on master? 16:24:11 <evgeniyl> aglarendil: Hmm, we want to update repos from astute? 16:24:45 <evgeniyl> aglarendil: why we can't do it from puppet when we start it? 16:25:19 <vkozhukalov> i see a couple review requests and one of them is already merged and it is really huge 16:25:24 <vkozhukalov> #link https://review.openstack.org/#/c/84092/ 16:25:29 <vkozhukalov> great 16:25:34 <mattymo> we could also set --environment variable and configure the paths for each versioned puppet manifests in puppet.conf 16:25:41 <mattymo> it's a little tricky too 16:25:50 <mattymo> and not any better than --modulepath= 16:25:51 <aglarendil> it is more convenient to do in astute 16:26:03 <aglarendil> because you are not able to change prefetched data in the runtime 16:26:56 <evgeniyl> aglarendil: don't agree, but ok, we can discuss it later 16:27:26 <vkozhukalov> later in ML 16:27:31 <vkozhukalov> moving on 16:27:57 <vkozhukalov> #link https://blueprints.launchpad.net/fuel/+spec/relocate-haproxy-to-its-own-network-namespace 16:28:28 <aglarendil> implemented 16:28:32 <aglarendil> and merged 16:28:45 <xarses> yay! 16:28:45 <vkozhukalov> great 16:28:54 <xarses> so ha works now? 16:28:55 <mihgen> verified? 16:29:10 <aglarendil> It works - means it is verified :) 16:29:12 <vkozhukalov> looks like we got rid of problems with reconnecting 16:29:14 <mihgen> ha worked before .. 16:29:18 <mihgen> but not ideally ;) 16:29:32 <vkozhukalov> moving on 16:29:53 <vkozhukalov> #topic https://blueprints.launchpad.net/fuel/+spec/nailgun-objects-flow 16:30:09 <meow-nofer> currently almost all major items we have are rewritten to objects, so they implement the same interface 16:30:25 <meow-nofer> that's quite enough for current API freeze 16:30:49 <meow-nofer> Node object is already merged and Notification object (small and simple) is on the way 16:30:51 <mihgen> do we have any versioning of RPC now with it? 16:31:13 <meow-nofer> not yet, but is's not really hard to implement 16:31:14 <xarses> meow-nofer: are we going to fix the import issue on NetworkManager? "from nailgun.network.manager import NetworkManager" 16:31:19 <evgeniyl> mihgen: it's a part of patching openstack feature 16:31:27 <mihgen> xarses: it's fixed I believe 16:31:31 <evgeniyl> mihgen: i.e. no, we on't have versioning for rpc 16:31:36 <meow-nofer> xarses, already fixed https://review.openstack.org/#/c/84739/ 16:31:48 <xarses> meow-nofer: thanks 16:32:17 <meow-nofer> RPC versioning is not the same as REST versioning, let's discuss it separately later 16:32:22 <meow-nofer> let's move on 16:32:24 <vkozhukalov> #link https://blueprints.launchpad.net/fuel/+spec/nailgun-rest-db-constraint 16:33:04 <meow-nofer> Igor implemented simple wrapper, so now you if you get 500 error from API - it is something really strange 16:33:46 <meow-nofer> like, modifying objects if you're breaking some constraints using API is now wrapped and will never return 500 16:33:54 <meow-nofer> it's 400 as it should be 16:34:10 <meow-nofer> any questions on this one? 16:34:26 <meow-nofer> if no, let's move 16:35:00 <vkozhukalov> ok, thanks 16:35:17 <vkozhukalov> #link https://blueprints.launchpad.net/fuel/+spec/nailgun-rest-api-pagination 16:35:37 <vkozhukalov> looks like it is important for notifications 16:35:50 <vkozhukalov> only 16:35:57 <meow-nofer> work on pagination is already in progress and it became much simpler after we introduced objects for all major entities 16:35:59 <mihgen> for large installations I think 16:36:11 <mihgen> 500+ nodes 16:36:22 <evgeniyl> I think we need to change "Implementation" status. 16:36:29 <meow-nofer> so, on UIright now it's really important for notifications and we're discussing how to use it with Nodes 16:37:15 <meow-nofer> I don't think we'll implement this for nodes on UI in this release, though 16:37:24 <meow-nofer> for nodes, I mean 16:37:38 <evgeniyl> It will be not so easy to implement on UI, because we have grouping 16:37:46 <meow-nofer> yep 16:38:41 <vk> yep, notifications are great start 16:38:51 <vk> finially we will implement them properly 16:38:57 <vkozhukalov> moving on? 16:38:58 <meow-nofer> yay 16:39:02 <meow-nofer> go go 16:39:18 <vkozhukalov> #link https://blueprints.launchpad.net/fuel/+spec/nailgun-patch-method-implementation 16:39:24 <mihgen> go go night clubbing 16:39:31 <meow-nofer> go go power rangers 16:39:31 <evgeniyl> anotchenko: is working on it 16:39:51 <vkozhukalov> how is the progress? 16:39:52 <anotchenko> We already discussed implementation details, working on it right now 16:40:02 <vkozhukalov> great 16:40:13 <anotchenko> how to update nested structures 16:40:17 <meow-nofer> it's the last major thing we're planning to implement in this release 16:40:21 <anotchenko> and other details 16:40:21 <mihgen> will it be done soon? 16:40:25 <meow-nofer> dealing with REST API 16:40:43 <xenolog> guys, after moving Haproxy to it's network namespace, I think, we should remove read_timeout=60 from any mysql urls. 16:40:43 <xenolog> I tested it, but anyone another should test it. 16:40:52 <anotchenko> Well, review request by monday I think 16:41:07 <xarses> meow-nofer: so will we support both PATCH and PUT or only PUT? 16:41:11 <meow-nofer> xenolog, offtopic, write about it on fuel-dev 16:41:29 <anotchenko> There not so much uses for PUT, but in some cases , probably 16:41:42 <meow-nofer> xarses, only PATCH, I would say, because right now our PUT is working exactly as PATCH should 16:41:58 <meow-nofer> but maybe we'll leave PUT in some places, too 16:42:25 <vkozhukalov> ok 16:42:32 <xarses> meow-nofer: ok 16:42:33 <meow-nofer> k 16:42:44 <meow-nofer> ) 16:42:52 <vkozhukalov> #link https://blueprints.launchpad.net/fuel/+spec/fuelclient-refactoring 16:43:26 <vkozhukalov> review request 16:43:27 <anotchenko> reviewed by two people, 5-th patchset https://review.openstack.org/#/c/83693/5 16:43:43 <anotchenko> Fixing latest comments 16:43:48 <vkozhukalov> and thousands of lines ) 16:43:56 <anotchenko> +3000, -2000 16:44:11 <xarses> it already was 2k lines 16:44:11 <meow-nofer> it looks like a big arguable thing, but I looked through it and it requires a lot more comments, at least 16:44:45 <meow-nofer> please, colleagues, let's review this more thoroughly 16:44:55 <anotchenko> meow-nofer: I’ll add some comments in next patchset 16:45:06 <vkozhukalov> ok, waiting for comments and additional review 16:45:22 <xarses> meow-nofer: rmoe and I have some thoughts about fuelclient, we will look it over too 16:45:46 <xarses> I see commit message needs link to bp 16:46:03 <meow-nofer> xarses, the main thing is, we looked through existing solutions 16:46:32 <meow-nofer> xarses, and they didn't look really suitable and simple in some cases 16:46:53 <meow-nofer> xarses, so, anotchenko did write an implementation itself, that's why it is so big 16:46:59 <xarses> meow-nofer: any change in this direction is helpful. 16:47:38 <meow-nofer> xarses, yes, they use some thing called cliff in openstack 16:47:43 <anotchenko> It’s big cause it has structure and a lot more licence copypaste 16:47:58 <vkozhukalov> ok, will discuss in review request and maybe in ML 16:48:04 <vkozhukalov> moving on 16:48:50 <vkozhukalov> #topic Open Discussion 16:49:48 <mihgen> some folks participated in networking discussions today 16:49:59 <mihgen> I'm wondering what would be the first items to address 16:50:16 <mihgen> for network flexibitliy and large datacenters 16:50:37 <mihgen> actually xarses and rmoe should work on multiple l3 domains, wondering on status 16:51:44 <mihgen> take a look at this networking diagram: https://docs.google.com/a/mirantis.com/presentation/d/115vCujjWoQ0cLKgVclV59_y1sLDhn2zwjxEDmLYsTzI/edit#slide=id.g1ddc96bdd_071 16:52:02 <xarses> rmoe was working on cleaning up some of the last failing test cases that already exist. I think at that point we will ask to review the commit and add tests for new methods shortly after so that more people can look over the code 16:52:06 <mihgen> it's something which is definitely needed for large DCs 16:52:40 <mihgen> xarses: cool. So it's gonna be available in Fuel CLI, right? 16:53:07 <xarses> mihgen: after the fuelcli refactor, of course 16:53:30 <mihgen> we were discussing how it can be done in UI, as well as many other things for networking. vk is gonna put this down and share... 16:53:39 <mihgen> xarses: ohh there is dependency 16:53:44 <mihgen> do we have a bug/bp for it? 16:54:17 <xarses> we should update it somehow to show that, yes 16:54:20 <mihgen> or you gonna simply extend CLI as part of the larger work? 16:54:28 <xarses> the second 16:54:56 <mihgen> ok. you've heard that there is some work in parallel by anotchenko , refactoring stuff… 16:55:00 <xarses> we where hoping for the fuelcli rework to land first so it would be easier to add the commands 16:55:00 <mihgen> I hope you are in sync 16:55:23 <xarses> yep 16:55:41 <mihgen> cool 16:56:16 <anotchenko> xarses: mihgen : review’ll be over soon 16:56:35 <xarses> anotchenko: =) 16:57:05 <mihgen> any other topic to discuss now? 16:57:12 <vkozhukalov> no 16:57:25 <vkozhukalov> ok, thank you all 16:57:29 <vkozhukalov> great meeting 16:57:37 <evgeniyl> thanks 16:57:51 <mihgen> have a great night / day ;) 16:58:02 <meow-nofer> thx by xoxo 16:58:19 <vkozhukalov> #endmeeting Fuel