13:00:50 <yanyanhu> #startmeeting senlin 13:00:51 <openstack> Meeting started Tue Jan 24 13:00:50 2017 UTC and is due to finish in 60 minutes. The chair is yanyanhu. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:53 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:56 <openstack> The meeting name has been set to 'senlin' 13:01:04 <yanyanhu> hi, guys 13:01:09 <Qiming> lhi 13:01:21 <XueFeng> hi,yanyan, QiMing 13:01:26 <lxinhui> hi 13:01:28 <yanyanhu> hi, Qiming XueFeng 13:01:31 <yanyanhu> evening 13:01:35 <elynn> Hi 13:01:37 <XueFeng> hi xinhui 13:01:46 <XueFeng> hi elynn 13:01:46 <yanyanhu> hi, elynn 13:02:32 <yanyanhu> ok, lets get started 13:02:45 <yanyanhu> here is the agenda, please feel free to add items you want to discuss 13:02:46 <yanyanhu> https://wiki.openstack.org/wiki/Meetings/SenlinAgenda#Agenda_.282017-01-24_1300_UTC.29 13:03:05 <yanyanhu> #topic Ocata-3 release 13:03:24 <yanyanhu> this week is the o3 release 13:03:44 <yanyanhu> so we will cut o3 release of senlin service 13:03:53 <Qiming> there seem to be quite a few patches which will need sdk 0.9.12 for release 13:03:54 <yanyanhu> and also make the final release of senlinclient 13:04:11 <yanyanhu> Qiming: yes, Steve has updated the patch to bump sdk version 13:04:13 <Qiming> need to closely track the sdk version bump patch 13:04:20 <yanyanhu> hopefully it can be done in one or two days 13:04:24 <yanyanhu> Qiming: yes 13:04:25 <Qiming> yes 13:04:36 <yanyanhu> https://review.openstack.org/#/c/418663/ 13:04:53 <yanyanhu> after it is done, we can safely but senlinclient release :) 13:05:24 <yanyanhu> about senlin service, is there any critical patch is still pending? 13:05:36 <Qiming> yep, there are quite a few patches depending on the new sdk release 13:05:42 <yanyanhu> https://review.openstack.org/#/q/project:openstack/senlin+status:open 13:06:31 <yanyanhu> I guess those fixes about health manager should be included 13:07:00 <yanyanhu> for mistral driver, it is still in progess I think 13:07:06 <yanyanhu> and also the vdu profile 13:07:30 <Qiming> yes 13:07:46 <yanyanhu> I see, will stare at them 13:07:55 <yanyanhu> and please just leave message or ping me if you have any critical patches need to deal with :) 13:08:16 <XueFeng> ok 13:08:20 <yanyanhu> I plan to cut the release by the end of Thursday 13:08:30 <yanyanhu> for both service and client 13:09:02 <yanyanhu> great, if no other questions, lets move on 13:09:16 <yanyanhu> #topic Proposals for Boston Summit 13:09:21 <yanyanhu> https://etherpad.openstack.org/p/senlin-boston-summit-proposal 13:09:40 <yanyanhu> we now have three proposals in the list 13:09:53 <yanyanhu> two for NFV related topics and one for dtdream use case 13:10:14 <yanyanhu> and both of them have a etherpad to collect the idea 13:10:27 <yanyanhu> hope you guys can take a look at them and leave your comments 13:10:36 <yanyanhu> since the deadline is 6th, Feb 13:10:53 <yanyanhu> we may need to finish the proposal during our Spring festival :) 13:11:40 <yanyanhu> hi, lxinhui, elynn, any thing you guys want to share :) 13:11:57 <yanyanhu> I guess haiwei is not here 13:12:05 <elynn> I was preparing the edu profile 13:12:08 <yanyanhu> ruijie: hi, evening 13:12:18 <ruijie> hi yanyanhu 13:12:24 <lxinhui> Ethan and bran are working together 13:12:27 <elynn> and also add needed attributes to heat senile resource 13:12:31 <yanyanhu> elynn: great, that is for supporting the demo(if we have one) 13:12:37 <lxinhui> on the VDU dev and tests 13:12:55 <yanyanhu> lxinhui: nice 13:13:02 <lxinhui> VMware has test bed will complete network support 13:13:15 <yanyanhu> so have you submit the proposal to the website of summit? 13:13:16 <lxinhui> will throw out draft for your comments 13:13:31 <lxinhui> not yet 13:13:40 <yanyanhu> great 13:13:46 <yanyanhu> will help to review it 13:13:52 <lxinhui> thanks 13:13:57 <yanyanhu> hi, ruijie, you are on the train? 13:14:08 <ruijie> Not yet :) 13:14:16 <yanyanhu> :) 13:14:47 <yanyanhu> will also review the proposal you leave in the etherpad and leave comments 13:15:02 <yanyanhu> hi, team, please help to review those proposals as well :) 13:15:19 <ruijie> Basically it's about ha and scalability for PaaS 13:15:22 <yanyanhu> hope we can get chance to show people some pratical use cases 13:15:37 <yanyanhu> ruijie: about pass, you mean? 13:15:55 <ruijie> Use senlin to manage our PaaS 13:16:02 <ruijie> The App platform 13:16:20 <yanyanhu> so the applications are also managed by senlin? 13:16:36 <ruijie> To provide HA and scalability for the VMs 13:16:50 <ruijie> Which contains a number of containers 13:17:17 <yanyanhu> have you tried senlin's container cluster :P 13:17:21 <ruijie> The app are managed by the cloud foundry 13:17:27 <yanyanhu> I see 13:17:51 <yanyanhu> so senlin is for preparing the iaas and cloudfoundry is staying upon it 13:18:06 <ruijie> Yes yanyanhu 13:18:15 <yanyanhu> nice 13:18:23 <ruijie> IaaS-senlin-cloud foundry 13:19:02 <yanyanhu> if so, I guess senlin can provide the ability of scaling and HA for cloudfoundry cluster 13:19:16 <ruijie> But I am not sure it's specific since ha and scalability has been talked for a long time 13:19:42 <Qiming> yes, it is a huge field by itself 13:19:58 <ruijie> Right, and we can use ansible to solve the deployment problem 13:19:58 <Qiming> need to be more specific about the requirements 13:20:04 <yanyanhu> ruijie: yes, they are. But I think we can focus on our use scenario 13:20:47 <yanyanhu> and how it benefits from senlin's scaling and HA support 13:21:07 <ruijie> E.g we have 10 vms 13:21:22 <ruijie> It's a fixed number 13:21:44 <ruijie> Currently we use these 10 VM to provide scalability for apps 13:22:01 <ruijie> But you see the VMs number is fixed 13:22:19 <ruijie> If the load is high , it's not scalable at all 13:22:41 <ruijie> So we need 2 level scalability, VM and container 13:22:53 <yanyanhu> ruijie: yes 13:23:17 <yanyanhu> actually we need to align the scaling of VMs and Containers to achieve the scaling of entire system 13:23:37 <ruijie> And also, the VM should be able to be rebuilt or recreate if it goes down 13:23:47 <ruijie> Right 13:24:04 <Qiming> ruijie, we talked about that before, back on the Tokyo summit 13:24:32 <ruijie> Yes Qiming, I see, so I am not sure it's good 13:24:38 <Qiming> http://www.slideshare.net/cliffton75/autoscaling-with-magnum-and-senlin 13:24:57 <Qiming> it was a joint work with some university studentds 13:25:03 <ruijie> Just change magnum to cloud foundry 13:25:08 <Qiming> they got it done in one month 13:25:29 <Qiming> the devil is in the details 13:25:57 <Qiming> the metrics you collect, the way you collect those metrics, the threshold you set, the triggering mechanism, the cooldown period ... 13:26:41 <Qiming> when do you want to scale-in, when do you want to scale-out ... etc. 13:27:07 <yanyanhu> so maybe we consider to talk about it with a real use case in dtdream? 13:27:13 <ruijie> Yes Qiming, that's how we use it too... 13:27:35 <ruijie> Like Ali ESS 13:27:47 <yanyanhu> what application you are running, and what its requirement on scaling/HA, and how to use senlin to suppor it? 13:28:23 <Qiming> I'm not sure senlin can provide a complete solution, but senlin can evolve into a competent component for your integration 13:28:31 <yanyanhu> and some traps you met and some data for sharing? 13:28:41 <ruijie> SENLIN+ansible 13:28:48 <ruijie> To deployment 13:28:49 <yanyanhu> yep, senlin is like the bridge here 13:29:23 <Qiming> that was my vision a long time ago, I mean, senlin + ansible for clustered app/service deployment and operation 13:29:35 <yanyanhu> it will be a good story if you can build a complete workflow 13:30:33 <ruijie> Use senlin to create a cluster and cluster collect... ansible to deploy 13:30:52 <ruijie> This could create a cluster contains control nodes 13:31:16 <ruijie> And another is cluster contains compute resources 13:31:53 <ruijie> Ha policy scaling policy ... to provide HA and scalability for the compute cluster 13:32:58 <yanyanhu> so I feel describing the entire workflow clearly is important, and how it can be used to support your production environment 13:33:44 <Qiming> yep 13:35:01 <ruijie> Just we extend the profile plugin to run ansible workflow 13:35:07 <yanyanhu> so ruijie please consider to formulate the proposal from this aspect :) 13:35:27 <ruijie> Sure yanyanhu 13:35:40 <yanyanhu> ruijie: that's is not a problem, you can introduce that in the presentation :) 13:36:11 <yanyanhu> great 13:36:16 <ruijie> Thanks, will think about it :) 13:36:22 <yanyanhu> thanks :) 13:36:46 <yanyanhu> so any more question about this topic? 13:36:59 <ruijie> :) 13:37:03 <yanyanhu> please just notice the deadline :P 13:37:16 <ruijie> Feb 6th 13:37:22 <yanyanhu> yep 13:37:32 <yanyanhu> ok, lets move on 13:37:53 <yanyanhu> #topic NFV related work item 13:38:04 <yanyanhu> lxinhui and elynn has introduced it 13:38:41 <lxinhui> yes 13:38:47 <yanyanhu> so please help to review the code, guys 13:38:55 <lxinhui> and the proposal 13:39:02 <yanyanhu> yep 13:39:08 <lxinhui> :) 13:39:10 <elynn> And the patch at heat side :) 13:39:20 <lxinhui> Yes :) 13:39:39 <yanyanhu> and lxinhui elynn, please feel free to start the discussion if you meet any critical issue need to address 13:40:04 <yanyanhu> especially some critical changes in senlin 13:40:20 <yanyanhu> lets figure it out together 13:40:25 <elynn> I will 13:40:33 <Qiming> I'll try my best to stay online, if youyou don't cry 13:40:34 <lxinhui> sure 13:40:44 <lxinhui> haha 13:40:44 <yanyanhu> Qiming: :) 13:40:51 <yanyanhu> youyou should have the highest priority 13:41:01 <Qiming> she was crying just now ... 13:41:08 <lxinhui> :) 13:41:19 <yanyanhu> Qiming: you can leave for her now I think :) 13:41:31 <yanyanhu> #topic Open discussions 13:41:37 <yanyanhu> open discussion now 13:41:55 <yanyanhu> so you guys are still in working status :) 13:42:14 <yanyanhu> only two days left before the most important festival in China :P 13:42:22 <ruijie> I m on the way to train station:) 13:42:38 <Qiming> I'm playing with a Go version of senlin service 13:42:42 <yanyanhu> oh, I think the next irc meeting could be cancelled 13:42:44 <Qiming> database is using etcd 13:42:48 <yanyanhu> Qiming: cool! 13:43:09 <yanyanhu> for most of the team will be in vacation 13:43:12 <Qiming> it is fun, a lot of things have to be done manually, for lacking of package support 13:43:38 <yanyanhu> Qiming: yes, python's lib is more abundon than go's I guess 13:44:00 <yanyanhu> there is even no built-in lib for random string generation... 13:44:09 <ruijie> But go could build once run anywhere :) 13:44:10 <yanyanhu> uuid 13:44:15 <yanyanhu> sorry, not random string 13:44:22 <Qiming> there is 13:44:46 <yanyanhu> oops, last time I tried to find one but failed... 13:44:58 <Qiming> oh, I am using a lbi 13:44:58 <yanyanhu> there is 3rd part lib for it 13:44:59 <Qiming> lib 13:45:08 <Qiming> github.com/pborman/uuid 13:45:14 <yanyanhu> yep, that oe :) 13:45:15 <yanyanhu> one 13:45:25 <Qiming> I'm using this 13:45:51 <Qiming> 47 p.CreatedAt := time.Now().Format("2006-01-02T15:04:05 MST") 13:46:05 <yanyanhu> :) 13:46:07 <elynn> You can start a company if go version of senile is finished ;) 13:46:12 <Qiming> this line of code cost me a few hours as well 13:46:16 <yanyanhu> definitely, haha 13:46:24 <Qiming> elynn, it is senlin, not senile ... 13:46:33 <yanyanhu> LoL 13:46:39 <elynn> my input method always correct me... 13:46:47 <elynn> In a wrong way... 13:46:48 <yanyanhu> :P 13:47:00 <yanyanhu> that means we are not that popular to become a new english word 13:47:10 <yanyanhu> :D 13:47:28 <yanyanhu> ok, so I guess we can end the meeting a little earlier today 13:47:38 <yanyanhu> actually I have another meeting in 10:00pm... 13:47:43 <yanyanhu> may need to leave for it 13:48:12 <yanyanhu> so please keep in touch in coming week :) 13:48:20 <Qiming> XueFeng, online? 13:48:26 <yanyanhu> and happy Chinese new year in advance 13:48:49 <elynn> Happy new year to everyone :) 13:48:55 <Qiming> guys, just want to bring up an issue 13:49:05 <XueFeng> hi Qiming 13:49:20 <Qiming> there are cases people build a chain of patches having one depending on another 13:49:26 <yanyanhu> hi, Qiming, so lets backup each other for the o3 relase, in case the network in the home of my parents in law is not good... 13:49:49 <Qiming> e.g. p3 is a branch out of p2 which is a branch from master 13:49:56 <yanyanhu> I will try to cut the release in Thursday 13:50:18 <Qiming> in those cases and other cases, please don't use the gerrit Edit functionality to change the code 13:50:22 <yanyanhu> thanks a lot 13:50:42 <Qiming> it will create a patch with a change-id only known to the server, now on the committer's local git repo 13:50:43 <yanyanhu> agree with Qiming, public edit in gerrit is not recommened 13:50:44 <elynn> thanks for remind 13:50:49 <yanyanhu> especially for cod 13:50:53 <yanyanhu> code 13:50:57 <Qiming> it will generate a lot chaos for rebase 13:51:41 <Qiming> all comments are welcomed, but please don't edit other's patch unless you know it is independent, already in good shape to be merged ... 13:51:51 <yanyanhu> so have to leave to prepare for next meeting, please help to end the meeting at 10:00, thanks :) 13:51:56 <yanyanhu> talk to you guys later 13:52:07 <Qiming> okay 13:52:34 <XueFeng> Ok, If we edit and download this patch 13:52:52 <XueFeng> It will work again? 13:53:19 <Qiming> no, in my local git, I may have patch 2 and patch 3 depending on that patch 13:53:42 <Qiming> e.g. this is what I have done locally 13:53:52 <Qiming> git checkout -b fix-bug-1 13:53:56 <Qiming> git commit 13:53:57 <Qiming> git review 13:54:05 <Qiming> git checkout -b fix-bug-2 13:54:08 <Qiming> git commit 13:54:11 <Qiming> git review 13:54:34 <Qiming> now, if someone changed the first patch, I cannot easily rebase my second one onto it 13:54:42 <Qiming> locally, I don't have the new change-id 13:55:05 <Qiming> in some cases, I may have a 3rd, 4th patch depending on each other 13:55:48 <Qiming> if someone has made changes at server side, I have to rework all local patches, and most of the time, it is a huge pain 13:55:54 <Qiming> a lot of rebase errors will happen 13:56:08 <XueFeng> Yes, if edit fix-bug-1 and we download fix-bug-1. 13:56:16 <XueFeng> then in fix-bug-2 13:56:45 <XueFeng> git rebase -i fix-bug-1 13:56:58 <XueFeng> It will ok 13:57:05 <Qiming> sometimes it work, sometimes it simply don't 13:57:25 <Qiming> especially when I have fix-bug-3 and fix-bug-4 13:57:30 <Qiming> it is ANNOYING 13:58:03 <XueFeng> if so maybe it bug of git 13:58:23 <XueFeng> s/it/it's a 13:58:34 <Qiming> a better option is STOP editing other's patch unless you know there is no dependency chain at the author side 13:58:53 <XueFeng> Yes 13:58:56 <Qiming> and the edit is the LAST thing to fix so it can be merged first 13:59:03 <Qiming> s/first/fast 13:59:09 <Qiming> that's all 13:59:10 <Qiming> thanks 13:59:17 <Qiming> running out of time 13:59:29 <Qiming> best wishes, guys 13:59:32 <Qiming> #endmeeting 13:59:47 <Qiming> sigh 14:00:08 <elynn> need yanyanhu back to endmeeting? 14:00:21 <Qiming> yes 14:00:27 <XueFeng> HaHa 14:00:31 <Qiming> or he can #chairn one of us 14:01:56 <Qiming> meeting chair left, then no one can #endmeeting ? 14:02:31 <jpich> Someone from infra may be able to? 14:03:05 <yanyanhu> I'm here will end it 14:03:08 <Qiming> an awkward situ 14:03:12 <yanyanhu> #endmeeting