13:00:26 #startmeeting senlin 13:00:27 Meeting started Tue Jul 7 13:00:26 2015 UTC and is due to finish in 60 minutes. The chair is Qiming. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:28 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:31 The meeting name has been set to 'senlin' 13:00:37 hi, there 13:00:41 hi 13:00:45 hi 13:00:53 hello 13:01:03 #topic add topics to agenda 13:01:06 yes 13:01:13 #link https://wiki.openstack.org/wiki/Meetings/SenlinAgenda 13:02:26 the first thing I have in mind is about the talk proposal 13:02:36 #topic Tokyo summit talk proposal 13:03:02 we have a draft abstract reviewed and approved for submission 13:03:20 I have tried to do this after having collected info from some of you 13:03:32 nice, so the abstract will be proposed for final voting? 13:03:44 but ... unfortunately, I don't think the website is friendly to Chinese developers 13:03:55 yes, yanyanhu 13:03:58 what do you mean 13:04:02 the website? 13:04:07 so I have asked lkarm to do me a favor 13:04:34 haiwei, the talk submission site is not working from China I believe 13:04:45 oh 13:04:53 every time you save the presentation (abstract indeed), it is still empty 13:04:56 oh, you mean the blocking done by the 'great firewall' 13:05:12 I have just submitted the proposal. We can still edit things until July 15 though if something needs to change 13:05:15 maybe some APIs made from that website point to Google 13:05:26 thanks, lkarm 13:05:37 do you have a link ? 13:05:59 presenters please check if you have a cool photo there, :) 13:05:59 https://www.openstack.org/summit/tokyo-2015/call-for-speakers/manage/4187/summary 13:06:37 em, I'm getting this: "You cannot edit this presentation" 13:06:56 I think probably since I submitted it, I am the only one that can edit it? 13:07:16 maybe 13:07:22 Can you not see it at all? 13:07:43 I cannot see anything except the single line of error message 13:08:35 anyway, we can check it offline 13:08:38 I can see that page 13:08:48 Oh maybe you have to be listed as a speaker to see it 13:08:59 i got this message : You can't edit this presentation 13:09:27 I'm working with some Magnum guys on a new proposal about the interaction between Magnum and Senlin 13:09:43 haiwei, you are also on the speaker list 13:09:50 it would be focused on managing clusters of containers 13:09:54 so I guess you can also access it 13:09:54 haiwei, what email did you use to sign in. maybe i used a different one 13:10:14 I just logged in using my work email 13:10:25 xu-haiwei@mxw.nes.nec.co.jp this is different from what I currently use 13:10:46 ok, leave it later 13:10:59 the container topic 13:11:23 the container specific topic is still under development 13:11:39 we give you all an update when we come up with something solid 13:12:06 ok, this is will be an very attractive topic I believe 13:12:20 in case you cannot edit the current submission, please drop lkarm an email 13:12:24 that means senlin will create a container type node? 13:12:46 yes, haiwei, we are recruiting some university students to do that 13:13:12 the work has just get started, they are studying magnum and senlin 13:13:27 let's move on 13:13:37 #topic liberty-2 milestone targets 13:13:55 please check the etherpad page here: https://etherpad.openstack.org/p/senlin-liberty-workitems 13:14:14 we have listed the todo items there 13:14:22 all targeting liberty-2 13:14:59 most of the items are about test cases 13:15:14 about unit test, I think we have a good progress till now 13:15:39 we have many "missing test case" bugs filed, and guys are claiming them 13:15:45 great! 13:16:13 yes, I think all bugs about test cases have been claimed :) 13:16:46 hopefully, we can achieve a 90% test coverage by the end of this month 13:17:02 that will be great 13:17:10 yes, I think we have big chance ;) 13:17:22 I will be focusing on the complex scaling policy implementation in the following weeks 13:17:50 ok, cool 13:18:15 that is another important feature we want to support in liberty release 13:18:16 the docs seems nearly finished, Qiming 13:18:26 just realized that we don't have a single line of test case for python-senlinclient ... 13:19:11 so maybe we should put this work item in liberty3? 13:19:14 well ... the getting started docs work is still not finished 13:19:16 python-senlinclient should be done in L3? 13:19:56 s/should/may have to/ 13:20:10 about the doc, I suggest we merge the first version if it has been ready and then complete other parts step by step 13:20:21 since it is so important for other guys to understand Senlin 13:20:49 and provide a guide about how to contribute to this project 13:20:50 okay, the patch is here: https://review.openstack.org/#/c/198538/ 13:21:12 and this is what it will look like when finished: http://docs-draft.openstack.org/38/198538/5/check/gate-senlin-docs/f4890a7//doc/build/html/getting_started/index.html#how-to-use 13:22:05 the current patch contains a rough structure, some drafts about profile types, profiles, and clusters 13:22:17 I think that is big enough for a patch to review 13:22:46 will start a new patch to work on docs related to nodes, membership, bindings, policies, etc 13:22:55 nice 13:23:35 hopefully this doc can be finished by this week 13:24:22 hi, Qiming, about the wiki page 13:24:46 is it actually similar to the doc? 13:24:55 about the content 13:25:21 yanyanhu, no, I don't think we will maintain two copies of the same content in different places 13:25:32 ok 13:25:56 so the wiki should give a more brief description? 13:26:05 on wiki page, we may want to emphasis on the high level designs and even visions 13:26:47 wiki site is not something we can do good version control 13:27:16 ok, understand 13:27:31 I think we have covered the documentation topic 13:27:38 #topic open discussions 13:28:19 something to update to you all 13:28:36 yanyanhu and I visited a customer last week 13:28:57 showing them the HA work we have done before and the Senlin project we are working on 13:29:13 they are pretty interested in this work 13:29:23 good news 13:29:30 thats great 13:29:39 they are actually from NFV (Network Function Virtualization) domain 13:29:49 HA is very important a feature to them 13:30:16 something similar here, a company want to do auto-scaling without using heat, our job is their desire 13:30:19 so I'm invited to share what we have done and what we are doing to their world-wide community tomorrow 13:30:45 haiwei, that sounds great. 13:30:53 when talking to them, we are all ears 13:31:05 very good news 13:31:20 we need to listen to them, capture their real requirements 13:31:45 I will give a brief introduction to the openstack community team in our company :) 13:32:01 thanks, haiwei 13:32:27 yet another thing to update 13:32:44 we have talked with some core contributors from the Magnum community 13:33:07 that is why are are proposing another talk proposal to the summit 13:33:34 it is about clustering and scaling support to Magnum, at least the infrastructure layer 13:33:59 Kubernetes and swarm each has their own solutions to scale containers 13:34:34 the main question is about collecting container metrics to trigger autoscaling 13:34:44 and using LBaaS for load balancing 13:35:01 the Magnum guys said they don't want to use stack forge project , didn't they? 13:35:02 there are other topics such as scheduling support 13:35:26 haiwei, yes, that what we heard from Adrian 13:35:54 but, things may change at unexpected paces 13:36:01 metric collection will probably have to happen through monasca. not sure if you can collect application level metrics from ceilometer 13:36:12 so push senlin integrated is important 13:36:20 so does Magnum mostly want to use the policies that senlin provides? 13:36:36 jruano, I did tried to send some samples from inside a VM to ceilometer, there are tricks 13:37:05 the difficulty mainly comes from authentication, you need a user/password, you need to know where the auth_url is 13:37:24 or else, ceilometer cannot accept your request 13:37:46 lkarm, they are evaluating senlin to build their bays 13:38:05 their bays are actually clusters in Senlin 13:38:14 actually, IMHO, senlin might be a good way to support Magnum Bay 13:38:16 oh okay thats really cool 13:39:02 currently, they are using Heat templates to create resource groups, that is not very flexible 13:39:07 and its structure also exactly matches Senlin cluster 13:39:46 I did heard that they can use LBaaS for load-balancing, yanyanhu 13:39:55 that is a good news too 13:40:07 oh, sure 13:40:38 that's all I have to share 13:40:51 if they can use LBaaS as the loadbalancer of Magnum dockers, some obstacle should have been cleaned by them 13:41:22 yes, I think so 13:41:23 the only problem is ourself don't have enough bandwidth to cover this part 13:42:13 glad to see lixinhui is starting code reviews as well 13:42:17 we need more hands 13:42:29 yep :) 13:42:46 Qiming, about the problem we meet when sync with the global requirement, I saw you propse a new patch 13:42:58 yanyanhu, yes, working on that 13:43:01 i can help there... i am currently developing the resource support for magnum within heat 13:43:20 great! 13:43:41 jruano, I saw your patch, will jump onto it when it is finished 13:44:06 cool... i have all the code ready. just need some feedback on the current direction 13:44:38 yanyanhu, the requirements error was caused by the version of Routes package, need different environment in tox.ini to make the gate happy 13:44:55 about the patch, I think maybe we can copy requirement.txt to requirement-py3.txt and only change the line that describes the package version of routes 13:45:14 jruano, I will review when you patch is submitted 13:45:25 and let py34 testenv depends on this new requirement-py3.txt? 13:45:31 yanyanhu, tried that, it didn't work 13:45:36 ok... 13:45:56 just recalled that I saw a similar patch in other project, but can't remeber the detail 13:46:07 there is a new restriction, no duplicate items in two requirements file is allowed now 13:46:37 and pip is complaining when I tried different versions 13:46:48 still struggling to get things straight 13:47:20 I think I can work it out today 13:47:26 em, since this issue also blocked the version sync of oslo.db package 13:47:28 cool 13:47:53 yep, that's the reason I didn't approve the requirements change so far 13:48:05 yes :) 13:48:22 anything else? 13:48:41 another thing is about the splitting of cluster_action, maybe this is also a workitem in L3? 13:49:03 well ... 13:49:10 we may have to 13:49:43 it won't be a huge task to refactor the test cases then 13:49:51 yes 13:50:06 about the presentation proposal, lkarm, does it related to the mail? 13:50:09 feel free to add this to the etherpad 13:50:14 and jruano, maybe you can also split the test case of cluster_action into multiple ones since that could be a huge module :) 13:50:39 haiwai, yes i used a different email address than the one that you posted above. i can try to change it 13:50:43 Qiming, ok 13:50:57 sure, i am starting the test case implementation today so i will figure that out later today 13:51:00 thanks , lkarm 13:51:22 jruano_, great ;) 13:51:40 +1 13:51:57 any module larger than 2000 lines would be a headache 13:52:21 oh, just filed a bug about using oslo_utils.timeutils to replace datetime, everyone is welcome to claim it :) 13:52:54 haiwei can you try to access now? 13:53:15 * Qiming wonders he should change is subscription to bugs using a different mail address 13:53:17 lkarm, it is ok now 13:53:21 thanks 13:53:27 good 13:54:21 haiwei, when the vote is open, don't forget to call for help for your colleagues, ;) 13:54:45 of course, sir 13:54:59 just remind me 13:55:01 :) 13:55:10 I WILL 13:55:23 anything else? 13:55:33 nope from me 13:56:08 okay, thank you guys for joining 13:56:13 #endmeeting