20:00:36 #startmeeting state-management 20:00:38 Meeting started Thu Oct 3 20:00:36 2013 UTC and is due to finish in 60 minutes. The chair is harlowja. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:39 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:41 The meeting name has been set to 'state_management' 20:00:47 howdy! 20:00:52 anyone around today :-P 20:00:53 hi there 20:00:58 hey 20:01:17 here 20:02:11 howdy 20:02:19 * harlowja waits a few minutes for others 20:02:29 #link http://eavesdrop.openstack.org/meetings/state_management/2013/state_management.2013-09-26-19.59.html (last meeting) 20:02:42 #link https://wiki.openstack.org/wiki/Meetings/StateManagement#Agenda_for_next_meeting (this meeting) 20:05:32 ok dokie 20:05:33 anyays 20:05:38 guess its just us, so might be quick :-P 20:05:49 #topic action-items 20:05:52 i guess i just had one, ha 20:06:44 so Caitlin56 it doesn't appear u can cross link blueprints :( 20:06:49 *cross project link 20:07:14 So enter them twice and type the URL in yourself, I suppose. 20:07:34 or like create one, then create a 'link' one that in its description has a link to the other one :-/ 20:07:41 sorta not so nice 20:08:04 *especially for a project like openstack where alot of stuff is cross-linked 20:09:52 cool 20:10:01 #topic coordination 20:10:19 so i think there is some work going on with the celery stuff that could benefit from a little more coordiation 20:10:35 let me see if any of those guys are around, kebray yt 20:10:47 I'm here for like five minutes 20:10:54 : X 20:10:57 k 20:11:35 so i was thinking that if we get the real basics in, even with @celery.task decorator, then we can proceed forward with fixing that problem 20:11:44 does that seem reasonable, i'm not sure how easy it will be to fix that 20:11:59 *the dependence on celery.task and ultimately its static registry 20:12:05 Yeah, I'm okay with that 20:12:13 Not sure what changes I need to make for that to happen though 20:12:22 Just pull out the event thread to consumer stuff? 20:12:30 s/pull out/change 20:12:30 sure 20:12:47 i think thats fair, and i think we need to writeup what the issue is, and have a little plan to address it 20:12:54 Will that change let distributed in, I guess was the question 20:12:56 which i think melnikov, u, and me can explore 20:13:18 sure, that change, + bugs/blueprints/plan on next steps to fix that would be great 20:13:23 i think thats reasonable 20:13:29 kk - Must go for school stuff, but I will get that doe 20:13:30 done * 20:13:37 * harlowja for those that have no idea what we are talking about, the writeup/plan would help 20:13:58 ok, so thats good 20:14:26 melnikov does that seem ok with u, i'd almost want to put the distributed_engine in some sorta of 'alpha' area if i could, due to these restrictions 20:15:08 but at least documenting bugs, blueprints, a plan to get out of the problematic situation would be a start 20:15:18 at least then we agree it is a problem and have some sort of path to fixing it 20:15:22 sounds reasonable 20:15:24 *instead of no path 20:15:41 *whether it can be fixed is another question, but we can talk about that outside of this meeting i gues 20:15:50 ok cool 20:16:09 #topic resumption-migrations 20:16:51 so melnikov and i have been throwing around ideas for how to migrate the persisted flow/task details and just wanted to make people aware (or get other input) 20:16:53 #link https://blueprints.launchpad.net/taskflow/+spec/resumption-migrations 20:16:59 thx melnikov for making that blueprint 20:17:10 there is a couple of prototypes up 20:17:46 the general idea though is that when u have a flow with a set of tasks running, while those tasks are running taskflow is actively persisting what those tasks have done 20:18:08 but now say half-way through running that set of tasks, the program is stopped, say for a software upgrade 20:18:40 then what happens when the program comes back online and needs to migrate the persisted data for those tasks to some newer format 20:18:48 so that the flow that was running can resume running 20:19:12 if anyone wants to comment on that blueprint with ideas, its fine also 20:19:29 a couple different code reviews with different ideas also 20:19:30 #link https://review.openstack.org/#/c/48773/ 20:19:35 #link https://review.openstack.org/#/c/49380/ 20:19:54 it'd be nice to see what other people think, are we makign it to complex, to simple, shouldn't do it at all (any of the above) 20:21:26 thats about all i had, seeking of course more input on designs and such :) 20:21:34 any other topics before open-discuss? 20:22:03 #topic open-discuss 20:22:12 so just to keep everyone updated 20:22:16 #link https://etherpad.openstack.org/TaskflowHKIdeas 20:22:25 also feel free to update that, still collecting ideas 20:22:51 melnikov i was thinking maybe i can focus on improving some examples and such, or maybe anastasia wants to, either or 20:22:54 or both of us can 20:23:06 get the resumption stuff in 20:23:30 and i can also start on whatever the talk will be about with kebray (who i think is going now!) 20:23:43 anastasia is on vacations until oct 14th 20:23:49 np 20:24:02 i can do examples then also, and whatever other stuff y! throws at me 20:24:30 maybe i can start a 'best-practices-for-taskflow' wiki also 20:24:33 that seems like a good idea 20:24:41 yes, good idea 20:24:52 +1 20:24:54 #action harlowja start taskflow-best-practices wiki 20:25:13 since i think we'll hit alot of the same questions over and over 20:25:26 and things that taskflow is good at, and thinks that it is not 20:25:41 so this seems like a good place to direct people to for common stuff like that 20:25:53 will do that 20:26:13 Caitlin56 is any of the nextenta people doing a session for cinder 20:26:16 if u know 20:26:33 ah, i see http://summit.openstack.org/cfp/details/70 20:26:40 cool 20:27:18 planned but not formally scheduled as of yet, 20:27:21 sure sure 20:27:41 most of those seem 'Unreviewed' still 20:27:53 I wouldn't be opposed to presenting a version in a second session if you'd like. 20:27:57 should start transitioning to 'approved' (or not) hopefully i guess in the next couple of weeks 20:28:30 jgriffith is fighting havana bugs - not thinking about icehouse yet 20:28:38 either or, i was just thinking u might have some good ideas for https://etherpad.openstack.org/TaskflowHKIdeas 20:28:44 and we can combine our idea power! 20:28:45 ha 20:29:26 and if i can help out with your session, let me know also 20:29:34 But I am finally confirmed for Hong Kong. 20:29:41 thats step #1 i guess ;) 20:29:48 harlowja hey, I'm here, but dealing with an internal issue... so, not really here. 20:29:51 np 20:30:01 kebray jessica popped in, we i think have a good plan 20:30:19 just i need to connect with u on talk and stuff 20:30:32 can do after/outside meeting 20:32:03 ok, cool, any other stuff for people to talk about, or we can end early :) 20:32:09 speak now!, ha 20:34:17 alright then 20:34:34 for more stuff #openstack-state-management folks! 20:34:40 u'll find most of us in there 20:36:46 thx for coming folks melnikov Caitlin56 ;) 20:36:56 #endmeeting