13:00:12 <ruijie_> #startmeeting senlin
13:00:13 <openstack> Meeting started Tue Sep 12 13:00:12 2017 UTC and is due to finish in 60 minutes.  The chair is ruijie_. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:14 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:00:17 <openstack> The meeting name has been set to 'senlin'
13:00:35 <ruijie_> here is the agenda: https://wiki.openstack.org/wiki/Meetings/SenlinAgenda#Agenda_.282017-09-12_1300_UTC.29
13:00:46 <ruijie_> feel free to add items if needed :)
13:01:25 <Qiming> hi, I have added two items
13:01:44 <ruijie_> yes Qiming, noticed that :)
13:02:02 <ruijie_> let's wait for a while ?
13:04:05 <ruijie_> well :) let's get started from the work items
13:04:34 <ruijie_> I just moved the work items to a new etherpad page, Qiming
13:05:48 <ruijie_> hi Qiming, are there updates about the node adoption
13:06:39 <Qiming> so that
13:06:50 <Qiming> no
13:07:49 <ruijie_> okay Qiming, I am going to finish the gc work this week. And will look back to the scaling improvement :)
13:08:12 <Qiming> ok
13:09:19 <ruijie_> em, as I can see, there are no critical bugs left
13:11:40 <ruijie_> next one is about the "api requests with extra field"
13:11:47 <ruijie_> Qiming, your tern :)
13:12:12 <Qiming> alright, got request this morning
13:12:13 <ruijie_> s/tern/turn
13:12:23 <Qiming> it was on #senlin channel
13:12:44 <Qiming> someone asked if there are ways to intercept a scaling operation
13:13:13 <Qiming> for example, before a cluster is scaled in, we invoke a hook
13:13:27 <Qiming> the user wanted to do something
13:13:47 <Qiming> then confirm the scale in operation if it is permitted
13:14:52 <ruijie_> http://docs.aws.amazon.com/autoscaling/latest/userguide/lifecycle-hooks.html?
13:15:46 <ruijie_> it's about we can intercepting the scaling actions and then roll back ?
13:16:08 <Qiming> it doesn't have to be roll back
13:16:31 <Qiming> just leave the users some options
13:16:40 <Qiming> I can imagine a simple policy to do that
13:17:31 <Qiming> the real challenge is about how can we differentiate the original request which needed to be intercepted and  the resent one, which meant to be performed anyway
13:18:35 <ruijie_> like threetee said, send a signal to the VM to ask the app on it to shutdown, and after that the VM will send an notification so that we can really destroy the VM
13:19:13 <Qiming> if it is only about a signal
13:19:22 <Qiming> it can be done easily using a policy
13:19:55 <Qiming> another similar request we go early is about scaling out
13:20:15 <Qiming> before scaling out, that user wants to be notified
13:20:39 <Qiming> they need a management approval before the scale-out operation is performed
13:21:05 <ruijie_> yes Qiming, like a  work order
13:21:14 <Qiming> that led me to the thought about extra API parameter
13:21:58 <Qiming> telling us where the request came from, or telling us whether it should skip a policy check
13:22:34 <Qiming> for example, a request comes with an 'extra' field.
13:23:01 <Qiming> if the scaling request comes from a metering/alarming service, that field would be empty, by default
13:23:39 <Qiming> the policy can then process the request and redirect it somewhere, calling a hook, trigger a business process service or a workflow ... whatever
13:24:35 <ruijie_> yes Qiming, we can do that to ask for aprovement
13:24:36 <Qiming> when that request is approved (maybe with some modifications to its parameters), the 'extra' field now has a field {'approved': true}
13:25:25 <Qiming> the policy looks at the request again, and refrain from redirecting it, go ahead and execute it as usual
13:25:56 <Qiming> the point is the second request has to be somehow different from the initial one
13:26:38 <Qiming> did I make a point?
13:26:46 <ruijie_> yes Qiming
13:27:04 <ruijie_> I thought we can do that before we create any actions?
13:27:27 <ruijie_> so that we don't need to care about how long this process will take or whether it will block out engine
13:27:27 <Qiming> em ... don't think so
13:28:03 <Qiming> policies are only checked when an action is about to be executed
13:28:23 <Qiming> unless we explicitly add a new concept, e.g. hook
13:28:49 <ruijie_> yes, a watcher..
13:28:50 <Qiming> then we have to decide to what scope a hook can cover
13:29:36 <ruijie_> hi XueFengLiu
13:29:49 <XueFengLiu> hi,ruijie_
13:30:05 <XueFengLiu> hi, all
13:31:04 <ruijie_> we are just talking about adding extra fields to api request. http://docs.aws.amazon.com/autoscaling/latest/userguide/lifecycle-hooks.html?
13:33:05 <XueFengLiu> looks good. Let me check the meeting history
13:35:46 <Qiming> alright, next thing is about event notification documentation
13:36:06 <Qiming> I have just realized that we don't have proper documentation for users
13:36:19 <Qiming> Looks to me we only have developer docs
13:36:47 <ruijie_> yes Qiming :)
13:37:48 <Qiming> some of the text here are really targeting users rather than developers
13:37:50 <Qiming> http://git.openstack.org/cgit/openstack/senlin/tree/doc/source/contributor/event_dispatcher.rst
13:38:47 <Qiming> the overview section and the last section are both for users
13:39:24 <ruijie_> we need a user reference and one for developer guidelines perpose ?
13:39:37 <Qiming> I think s
13:39:39 <Qiming> so
13:39:58 <Qiming> that is how the docs are organized today
13:40:20 <XueFengLiu> Yes
13:41:01 <Qiming> see here: https://docs.openstack.org/senlin/latest/
13:41:44 <Qiming> section 2-4 are for users, section 5 is for developers
13:41:56 <Qiming> the community may have some new guidelines
13:42:12 <XueFengLiu> Qiming
13:42:27 <Qiming> I'm just too tired to follow all those guidelines
13:42:46 <XueFengLiu> Some new user may can't know which is for user which is for developer
13:42:59 <Qiming> then they can read them all
13:43:24 <Qiming> section 5 has a title, right?
13:43:40 <Qiming> a user won't touch that section
13:44:12 <Qiming> all developers are supposed to read them all, because they are supposed to start as a user
13:44:24 <XueFengLiu> Yes
13:44:44 <Qiming> okay, ruijie_
13:44:53 <XueFengLiu> From the guideline, some section title is the same
13:44:54 <Qiming> I have got 3 other topics
13:45:00 <ruijie_> honestly, I didn't find the notification part on this page :)
13:45:04 <Qiming> can I quickly dump them here?
13:45:20 <Qiming> ruijie_, that is something we can improve
13:45:24 <ruijie_> sure Qiming
13:46:10 <Qiming> 1. pls help review https://review.openstack.org/#/c/500713/
13:46:23 <Qiming> and https://review.openstack.org/#/c/501548/
13:46:38 <Qiming> the developer pinged me asking why there are no reviews
13:46:55 <ruijie_> revieded +1, but did't got new patch yet :)
13:47:27 <Qiming> 2. as part of Queens community goal, we are about to separate tempest out
13:47:41 <XueFengLiu> OK, will review
13:47:51 <Qiming> I have created a repo on github.com, hopefully maintained all commit history
13:47:53 <Qiming> https://review.openstack.org/#/c/501548/
13:48:02 <Qiming> sorry, wrong link
13:48:07 <Qiming> https://github.com/tengqm/senlin-tempest-plugin
13:48:29 <Qiming> when it is ready, we are supposed to propose a new repo for senlin project to host the tests
13:48:48 <XueFengLiu> OK
13:49:34 <Qiming> 3. there are things insane in the community, as always, please pay attention if you get cycles
13:49:40 <Qiming> this one: https://review.openstack.org/#/c/502216/
13:49:50 <Qiming> it is reinventing senlin inside tacker
13:50:07 <Qiming> funny ... but no one can stop people doing that
13:50:26 <Qiming> this one: https://review.openstack.org/#/c/465782/
13:50:37 <Qiming> really pissed off
13:51:27 <Qiming> how could people be so arrogant?
13:51:52 <Qiming> that's all from my side, ruijie_
13:52:01 <Qiming> back to you
13:52:07 <ruijie_> so according to there understanding, we are manageing VMs and containers .. that is irresponsible
13:52:33 <ruijie_> okay Qiming, next one
13:52:42 <ruijie_> Queens Meetup :)
13:53:14 <ruijie_> we can find a place to have a meetup to discuss our goals for Queens
13:53:27 <Qiming> sure
13:53:48 <XueFengLiu> OK, got all. Will check all this points  tomorrow.Qiming, ruijie_
13:54:19 <ruijie_> or, since it is almost 10.01, we can choose 1. have a virtual meeting first, and then hold meetup in middle Oct 2. f2f directly :)
13:54:43 <XueFengLiu> +1
13:55:13 <Qiming> I'll be out of town during Oct holidays
13:55:44 <Qiming> mid Oct sounds good to me
13:56:13 <Qiming> I was reading that as " Oct 2."
13:56:29 <ruijie_> my bad :)
13:57:04 <ruijie_> okay Qiming, I am going to send a email to mailing list about the virtual meeting
13:57:13 <Qiming> it's your call
13:57:27 <ruijie_> and we can discuss the detail about the meetup later
13:57:40 <Qiming> ok
13:57:57 <ruijie_> any other topics to discuss :)
13:58:47 <Qiming> I'm looking at this: https://review.openstack.org/#/c/502757/
13:59:04 <Qiming> hoefully can figure out what went wrong
13:59:43 <ruijie_> https://review.openstack.org/#/c/502996/
13:59:56 <Qiming> ya
14:00:22 <ruijie_> timesout
14:00:26 <ruijie_> #endmeeting