20:00:26 <harlowja> #startmeeting state-management
20:00:27 <openstack> Meeting started Thu Oct 17 20:00:26 2013 UTC and is due to finish in 60 minutes.  The chair is harlowja. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:28 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:30 <openstack> The meeting name has been set to 'state_management'
20:00:39 <harlowja> howdy folks
20:00:47 <harlowja> state-management fun time starts nnooow
20:00:53 <harlowja> #link https://wiki.openstack.org/wiki/Meetings/StateManagement#Agenda_for_next_meeting
20:02:27 <harlowja> anyone here, ha
20:02:30 <harlowja> if not, short meeting :-P
20:02:40 <ativelkov_> Hi, I am here
20:02:46 <harlowja> hi hi
20:02:48 <gokrokve> Hi.
20:02:53 <melnikov> hi there
20:02:55 <tsufiev_> hello there
20:02:59 <harlowja> howdy
20:03:03 <harlowja> new people, sweet!
20:03:12 * harlowja will wait a few for others
20:04:30 <harlowja> alright :)
20:04:40 <harlowja> #topic last_time_action_items
20:04:55 <harlowja> #link http://eavesdrop.openstack.org/meetings/state_management/2013/state_management.2013-10-10-20.00.html
20:05:04 <harlowja> i think i had the action items so can describe results
20:05:17 <harlowja> one was from a previous week, and finally did it @ https://wiki.openstack.org/wiki/TaskFlow/Best_practices
20:05:26 <harlowja> some of what i consider good practices for taskflow usage
20:05:34 <harlowja> *feel free to tell me they aren't good, ha
20:06:19 <harlowja> melnikov do u want to add any, or do those sorta make sense?
20:06:27 <harlowja> *if u've checked them out
20:07:43 <harlowja> other things i was supposed to do
20:07:53 <harlowja> manila chit-chat about taskflow
20:07:53 <melnikov> i checked them out, looks nice
20:08:17 <harlowja> so i did manila chit-chat, gonna talk with one of there folks about taskflow (who is actively using/investigating it)
20:08:27 <harlowja> since manila is based off cinder, they'd like to follow some of the same patterns
20:08:38 <harlowja> thx melnikov
20:09:49 <harlowja> alright, don't think there was any other action items that i can remember
20:09:59 <harlowja> #topic overall-effort
20:10:24 <harlowja> so i think today is a pretty big milestone for us, 0.1 i think will be going onto pypi very shortly
20:10:40 <harlowja> its been like 4 months to get here, and i think we have a pretty good base to move forward with :)
20:10:52 <changbl> i still see a few commits in review?
20:11:04 <harlowja> ya, 2 i think
20:11:13 <harlowja> i don't expect the infra people to +2 for a day though
20:11:25 <ativelkov_> Sounds great. Is it related to Havanna release, or these are independent milestones?
20:11:26 <harlowja> #link https://review.openstack.org/#/c/52283/
20:11:36 <harlowja> #link https://review.openstack.org/#/c/52462/
20:11:49 <harlowja> ativelkov_ for now they are independent milestones, although i think its good to keep in a similar pace
20:12:30 <harlowja> so after i think 2 more reviews, melnikov do u want to approve, https://review.openstack.org/#/c/51338/
20:12:48 <harlowja> then when the above 2 reviews are infra approved, a package should show up on https://pypi.python.org/pypi/taskflow
20:12:56 <harlowja> *at least thats what the infra people told me would happen, ha
20:13:11 <harlowja> +- some additional voodoo, ha
20:14:32 <changbl> I guess commits like (https://review.openstack.org/#/c/52352/) will be left out from 0.1?
20:14:52 <harlowja> lets see, hopefully melnikov can get your comments addressed
20:15:08 <harlowja> i think its ok if it is, i don't think its critical
20:15:14 <changbl> k
20:16:09 <harlowja> so thanks to all that helped make it possible :)
20:16:55 <harlowja> *high five*
20:16:55 <harlowja> ha
20:17:03 <caitlin56> harlowja: I have some questions on the BestPractices page. Shift over to #state-management or just talk about them here?
20:17:03 <changbl> +1 to all
20:17:19 <harlowja> caitlin56 maybe in open-discuss we can do that?
20:17:22 <melnikov> +1
20:17:35 <melnikov> ^^ for high five)
20:17:41 <caitlin56> harlowja: sounds good.
20:17:53 <harlowja> coolness
20:18:30 <harlowja> so i hope by this weekend, we will have 0.1 package on pypi, all things working out, it seems reasonable
20:18:53 <harlowja> *unless i missed some voodoo that i have to do
20:19:02 <melnikov> i hoped to address those two storage things before release
20:19:24 <harlowja> if anyone knows i missed some voodoo, please let me know :)
20:19:24 <melnikov> should finish tomorrow
20:19:52 <harlowja> sounds good melnikov
20:20:05 <harlowja> #topic integration next steps for icehouse
20:20:30 <harlowja> so this one is an interesting one, and i think we have some good traction that likely will just have to wait for the summit to see whats next
20:20:44 <harlowja> although i'm all for trying to work with others before the summit, and i think we have some of that ongoing
20:20:52 <harlowja> nova @ https://etherpad.openstack.org/p/IcehouseConductorTasksNextSteps
20:21:09 <harlowja> cinder, https://etherpad.openstack.org/p/CinderTaskFlowFSM
20:21:10 <harlowja> and a few others
20:21:16 <harlowja> and lots of summit sessions
20:21:39 <harlowja> so i think we can just continue that, try to organize as much as we can, and then 'just make it happen'
20:22:13 <harlowja> maybe we should start a common icehouse etherpad, sound reasonable?
20:22:26 <harlowja> to keep track of what/where/who
20:22:42 <changbl> sounds good
20:23:03 <harlowja> k
20:23:08 <harlowja> #action harlowja start that etherpad
20:23:13 <melnikov> what about adding milestone to launchpad, to tag blueprints and stuff
20:23:26 <harlowja> agreed
20:23:35 <harlowja> melnikov do u want to try that, i think its not so hard
20:23:48 <changbl> what future milestones do we have now?
20:24:10 <harlowja> 0.2? :)
20:24:15 <changbl> haha
20:24:19 <changbl> contents?
20:24:31 <harlowja> i think we need to get the basic job stuff going, since thats missing
20:24:50 <harlowja> and maybe a basic locking api
20:25:11 <harlowja> or others on https://blueprints.launchpad.net/taskflow that people think are relevant
20:25:44 <harlowja> it'd be nice to have basic https://blueprints.launchpad.net/taskflow/+spec/conditional-flow-choices to
20:25:50 <harlowja> *etherpad there on whiteboard
20:25:56 <harlowja> and whatever else comes out of icehouse discussions
20:26:03 <harlowja> seem reasonable?
20:26:30 * harlowja we can adjust the priority of those blueprints if we want to
20:27:00 <gokrokve> When do you plan to release a version 0.2?
20:27:31 <melnikov> i thinking we need just one Icehouse milestone, and release 0.2 (0.3 and so on) as soon as we have enough major features ready
20:27:38 <melnikov> *thought
20:27:59 <changbl> Can someone illustrate more on conditional flow? seems we are making flows more complex
20:28:09 <harlowja> i'm fine with that, melnikov so at openstack milestones we would have a declared 'major' release, minor in between
20:28:11 <gokrokve> Sounds good. I am just thinking about other project that might depend on taskflow.
20:28:31 <gokrokve> Thay will not want to wait until Icehouse official release.
20:28:46 <harlowja> sure, then they can take in minor releases?
20:29:02 <ativelkov_> They can always use the latest trunk as well
20:29:12 * caitlin56 agrees with gokrokve. I want to add cinder code using taskflow by end of icehouse, not be ready to code for j*whatever.
20:30:31 <gokrokve> So you need to have 0.2 release tested and release at Icehouse M1 to allow others to use stable version for development and testing
20:30:32 <harlowja> caitlin56 i think thats fine, there will be a release, just i think what qualifies as 0.2 (0.3...) we can debate on, maybe each blueprint is a minor release
20:31:08 <caitlin56> we need an API freeze date well before the implementation freeze date.
20:31:31 <harlowja> sure, seems reasonable
20:31:37 <gokrokve> harlowja: You can do a release per feature but your VC tree will have a lot of branches for parallel deelopment.
20:32:07 <gokrokve> API freezy sounds very reasonable.
20:32:16 <gokrokve> API freeze sounds very reasonable.
20:32:34 <harlowja> ok, then maybe a good idea is for folks to check out https://blueprints.launchpad.net/taskflow and think about what they want in 0.2, the above was my suggestion :)
20:32:46 <harlowja> and i don't think those suggestions alter the API
20:33:07 <harlowja> then we can come back to this also after summit discussions
20:33:08 <gokrokve> I think you need to have some estimations for each BP and than plan release 0.2 accordingly taking into account dates and estimations.
20:33:22 <gokrokve> Don't forget to multiply ETA by 2. :-)
20:33:26 <harlowja> ha
20:33:43 <harlowja> also depends on who is going to do all those BPs ;)
20:33:49 <rakhmerov> or even 3.1415 :)
20:34:09 <harlowja> ya, release 0.31415 ftw, ha
20:34:39 <gokrokve> harlowja: this is multiplication factor Pi for estimations :-)
20:34:46 <harlowja> ah
20:34:49 <harlowja> haha
20:34:53 <rakhmerov> right :)
20:35:05 * caitlin56 isn't sure that you want to admit a given release is irrational.
20:35:11 <rakhmerov> kind of a joke, but it usually works pretty well from the experience
20:35:15 <harlowja> :)
20:35:42 <harlowja> ok, so how about the following, we can work on https://blueprints.launchpad.net/taskflow and expand and clear those up, so certain ones are more clear (addressing changbl question)
20:36:04 <harlowja> and we can then see about 3.145 estimations
20:36:10 <harlowja> and then come back here and fight over that, ha
20:36:20 <harlowja> *don't all take all the work at once, ha
20:36:25 <gokrokve> Sounds good.
20:36:29 <changbl> harlowja, please assign zk-logbook BP to me
20:36:32 <rakhmerov> good to me
20:36:33 <harlowja> sure
20:36:42 <harlowja> cool
20:37:10 <gokrokve> So lets agree that on next meeting we will have rough estimations for each BP.
20:37:12 <harlowja> and if some of them aren't clear, #openstack-state-management so that we can make them clear
20:37:17 <harlowja> i think thats fair gokrokve
20:37:31 <harlowja> *and pretty detailed summary of what it is
20:37:36 <gokrokve> By the way who is responsible for testing?
20:37:50 <harlowja> mr.jenkins
20:37:59 <harlowja> :)
20:38:14 <changbl> haha
20:38:21 <gokrokve> so u have only unit tests?
20:39:07 <harlowja> correct, but its a library, so we can pretty much tests the full system
20:39:09 <rakhmerov> I believe it doesn't assume anything else
20:39:20 <rakhmerov> yep
20:39:23 * harlowja not sure what a integration test would mean in this case
20:39:25 <caitlin56> You really need two distinct layers of testing. 1) test the engine. 2) test the patterns that use the engine.
20:39:30 <harlowja> ya, we have that
20:39:32 <rakhmerov> understandable
20:39:33 <gokrokve> ok. so we have ongoing testing with jenkins.
20:39:39 <harlowja> *mr.jenkins
20:39:43 <harlowja> :)
20:39:59 <harlowja> yes, on going mr.jenkins tests stuff
20:40:09 <harlowja> *not sure of the coverage, although we can probably get it
20:40:10 <rakhmerov> it's all, in fact, unit testing
20:40:14 <harlowja> i think we've been pretty dillegent about it
20:40:19 <gokrokve> What is the coverage?
20:40:46 <harlowja> unsure, haven't ran that, we should probably get an idea there
20:40:51 <gokrokve> Do you use static analysis? It should work great for library like code.
20:41:08 <harlowja> ya, openstack uses pylint and flake8 which are its static analysis
20:41:13 <caitlin56> The syntax allows specification of patterns that make no sense. So exhaustive testing is clearly not feasible.
20:41:15 <harlowja> *as static as u can get
20:41:36 <harlowja> gokrokve and mr.jenkins runs those static analysis for us
20:41:42 <gokrokve> ok.
20:41:42 <harlowja> mr.jenkins is very nice
20:41:49 <harlowja> i think we can get him to run coverage for us to
20:41:59 <harlowja> just likely haven't turned on that job
20:42:23 <harlowja> i can investigate that, shouldn't be hard
20:42:33 <gokrokve> Yes. It will be great. It is a good indicator of confidence.
20:42:40 <harlowja> sure
20:42:48 <harlowja> #action harlowja see about turning on coverage
20:43:01 <harlowja> i think its a small change to do that
20:43:02 <rakhmerov> but not 100% guarantee, imho
20:43:16 <harlowja> is anything a 100% guarantee :)
20:43:28 <harlowja> not even US government 100% guarantee
20:43:29 <harlowja> ha
20:43:29 <gokrokve> This is a real life. Nobody guarantee 100%.
20:43:29 <rakhmerov> ok, even 80%
20:43:30 <rakhmerov> :)
20:43:55 <changbl> +1 not even US government 100% guarantee :)
20:43:58 <harlowja> ;)
20:44:07 <rakhmerov> I mean, I've seen people making test coverage 100%
20:44:26 <harlowja> taskflow not recommended for missle/mars usage
20:44:26 <rakhmerov> buy the system kept working wrong
20:44:39 <harlowja> *my disclaimer, ha
20:44:40 <gokrokve> I saw projects with 100% coverage and great failures in production :-)
20:44:41 <rakhmerov> *but
20:44:51 <rakhmerov> exactly
20:45:01 <harlowja> if u running your missle system on taskflow, u might want to reconsider, ha
20:45:23 <rakhmerov> coverage is not a purpose itself, need to realize that
20:45:31 <harlowja> +1
20:45:36 <rakhmerov> not a goal evean
20:45:40 <rakhmerov> even
20:45:54 <harlowja> k, i'll get u guys that going
20:46:15 <harlowja> switching topic :)
20:46:20 <harlowja> #topic HK summit speaker ideas
20:46:26 <gokrokve> Yep. I remember a story when Soviet Lunar probe missed  Moon because of floatin number used in math library. Small errors produces a big deviations on large scale.
20:46:42 <rakhmerov> :)))) haha
20:46:42 <harlowja> ya, i think NASA did the same thing with mars
20:46:51 <harlowja> some metric conversion problem, lol
20:46:53 <rakhmerov> and not once :)))
20:47:19 <harlowja> #link https://etherpad.openstack.org/p/TaskflowHKIdeas
20:47:24 <harlowja> so i tried to fill in more of that one
20:47:39 <harlowja> *although currently the page isn't loading for me :-/
20:47:43 <harlowja> ah there we go
20:47:47 <rakhmerov> I have seen that
20:47:57 <harlowja> i'll likely have to slim it down
20:48:02 <harlowja> as i start making slides and stuff for the speaker stuff
20:48:06 <rakhmerov> good points, nothing actually to argue about
20:48:17 <harlowja> rakhmerov in openstack, always something to argue about :)
20:48:25 <harlowja> i bring my battle armor
20:48:30 <gokrokve> harlowja: Do you have a presentation on HK summit?
20:48:30 <rakhmerov> well, not in this case :)
20:48:38 <rakhmerov> seriously
20:48:46 <harlowja> gokrokve yup, not only presentation, complete speaker session
20:48:53 <rakhmerov> good point to bring up on the summit
20:48:54 <harlowja> where people ahve to listen to me and kebray talk
20:48:59 <rakhmerov> *points
20:49:34 <harlowja> #link http://openstacksummitnovember2013.sched.org/event/29f1f996b36aaf0febc5d43b6f53f2a4#.UmBNWSSoV-Q
20:49:34 <gokrokve> Great.
20:49:51 <harlowja> if u guys want to writeup some mistral stuff, i can try to include also
20:50:01 <harlowja> since Convection is mentioned in that overview
20:50:07 <harlowja> *aka mistral
20:50:26 <changbl> seems mistral is quite similar to what taskflow does?
20:50:38 <changbl> i read the blog from mirantis
20:50:38 <rakhmerov> well, I think we have things to discuss in this context
20:50:39 <gokrokve> harlowja: Sure.
20:50:55 <harlowja> changbl ongoing discussions there, is mistral just a API/service using taskflow...
20:51:00 <harlowja> idk quite yet either
20:51:08 <changbl> got it
20:51:15 <gokrokve> changbl: It is not a substitution of taskflow. it is a next layer around taskflow.
20:51:15 <changbl> we need to figure out
20:51:25 <harlowja> changbl of course
20:51:32 <harlowja> working with gokrokve rakhmerov (and others) on that
20:51:38 <harlowja> #openstack-mistral
20:51:43 <rakhmerov> basically, yes, that's right. Mistral was targeting to implement ideas in Convection
20:51:44 <rakhmerov> but
20:51:56 <rakhmerov> We have so many ideas above that :)))
20:52:03 <harlowja> lots of ideas :-P
20:52:12 <rakhmerov> right :)
20:52:23 <harlowja> so rakhmerov  gokrokve not only speaker session, but also summit sessions
20:52:24 <rakhmerov> dont' be angry at me for this :))
20:52:32 <changbl> rakhmerov, you are here, you wrote the blog:)
20:52:32 <harlowja> ha
20:52:34 <kebray> can we have a combined meeting with Mistral folks prior to summit?  We should figure out our similar and different views and at least know where we all stand prior to summit.
20:52:50 <harlowja> kebray i think thats a reasonable ask, i'd be up for that
20:52:51 <kebray> Anyone want to volunteer to pull that meeting together?  Can be over IRC, or google hangout, or whatever.
20:53:05 <rakhmerov> yeah, that's exactly what we're trying to do
20:53:09 <kebray> If not, I'll do it.. but, would prefer if someone else can drive it.
20:53:18 <harlowja> also i have http://summit.openstack.org/ (search for taskflow) so those are 2 different type of meetins
20:53:25 <harlowja> rakhmerov i think is workin on that
20:53:28 <rakhmerov> we have so many great things to discuss, you won't believe :)
20:53:32 <harlowja> a
20:53:34 <harlowja> *ha
20:53:42 <harlowja> *waiting for mind to be blown
20:53:51 <kebray> rakhmerov cool.. good to have more people involved!
20:54:06 <ativelkov_> yup, and it turns out that some of them require some beer to understand :-)
20:54:11 <rakhmerov> np, we're interested in that very much
20:54:21 <rakhmerov> :)) haha
20:54:32 <harlowja> rakhmerov so maybe when we decide a time, openstack-dev list, then others can join
20:54:36 <rakhmerov> recalling today's discussion, definitely
20:54:56 <rakhmerov> yes, sure
20:54:59 <harlowja> cool
20:55:28 <gokrokve> #AI Renat to organize a hangout meeting
20:55:47 <harlowja> #action Renat to organize a hangout meeting
20:55:53 <harlowja> one of those will work, ha
20:56:06 <harlowja> so i might skip a few of the agenda, for next time, not enough for  time left
20:56:17 <harlowja> seeing that 4 minutes left :)
20:56:24 <harlowja> #topic open-discuss
20:56:27 <gokrokve> #AI Gosha has to provide some text about Mistral for Josh
20:56:48 <harlowja> thx, that'd be suepr
20:56:59 <harlowja> *super
20:57:12 <harlowja> caitlin56 u around, 3 minutes, or maybe we can continue in #openstack-state-management
20:57:29 <harlowja> wanted to answer your question on the best practices i created
20:57:35 <rakhmerov> guys, I just wanna deliver the main message here: looks like we dont' have a service like this in the whole ecosystem. We need to carefully think about the target user group, project mission and main conceptual ideas
20:57:47 <harlowja> rakhmerov +1
20:57:50 <caitlin56> Can start. The key thing that is missing is a definitionof what a "Flow" is.
20:58:16 <harlowja> https://wiki.openstack.org/wiki/TaskFlow#Flows ?
20:58:17 <rakhmerov> the further is more interesting ;)
20:58:29 <caitlin56> Specifically, if you launch two tasks of the same pattern, what happens. Who is responsible for keeping the work separate?
20:58:32 <rakhmerov> yep, have seen that :)
20:59:14 <harlowja> caitlin56 so engines are what actually run the 'work'
20:59:38 <harlowja> lets continue in other channel i think
20:59:38 <rakhmerov> not a question for now. We need to define first what we want users (who exactly?) can do with this service
20:59:45 <caitlin56> Do the engines name all resources that the tasks work with? If so, highlight that in the best practices.
20:59:50 <rakhmerov> harlowja: agree
20:59:57 <harlowja> k
20:59:59 <harlowja> #endmeeting