20:00:42 #startmeeting heat 20:00:43 Meeting started Wed Oct 2 20:00:42 2013 UTC and is due to finish in 60 minutes. The chair is shardy. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:44 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:47 The meeting name has been set to 'heat' 20:00:52 #topic rollcall 20:00:54 o/ 20:00:59 \o 20:01:01 Hi all, who's around? 20:01:09 o/ 20:01:44 o/ 20:01:49 zaneb, jpeeler, therve, SpamapS? 20:01:55 hey i'm here 20:01:57 hi 20:02:00 I'm here too 20:02:22 Ok, let's get started, not much on the agenda today, maybe we can have a short meeting ;) 20:02:25 shardy: sick today, barely here 20:02:31 #topic Review last week's actions 20:02:44 Actually I don't think there were any: 20:02:54 nope 20:02:57 #link https://wiki.openstack.org/wiki/Meetings/HeatAgenda 20:03:12 #link http://eavesdrop.openstack.org/meetings/heat/2013/heat.2013-09-25-20.00.html 20:03:24 Hi 20:03:27 hello 20:03:43 Hi 20:03:51 yo 20:03:51 #topic RC1 status 20:04:05 So, branching for RC1 is imminent: 20:04:09 #link https://review.openstack.org/#/c/49384/ 20:04:11 yay! 20:04:18 did asalkeld have a chance this morning to look at that last lingering issue? 20:04:41 I woke up 7 minutes ago 20:04:47 morning :) 20:04:48 :-) 20:05:10 shardy: should we be reviewing that change? 20:05:16 asalkeld there is sa template attached to that bug now 20:05:22 ok 20:05:49 stevebaker: No, I just acked it since we're ready to branch, and it's a one line change bumping the version 20:06:06 #link https://launchpad.net/heat/+milestone/havana-rc1 20:06:21 So master will be open for Icehouse after than lands 20:06:54 however it would be really great if we can focus some effort on testing the Havana branch (ie RC1) over the next few days 20:07:08 as we have ~2weeks to catch any remaining issues which may require an RC2 20:07:09 sdake_ what is the bug with the attachment? 20:07:24 i dont have the bug # handy 20:07:31 ok, no worries 20:07:36 I'm awake 20:07:36 so fixes land in master and get backported to the release branch? 20:07:39 sorry, y'all 20:08:16 sdake_: So asalkeld fixed bug #1233882, but the Ref issues may or may not be an issue, we haven't yet confirmed 20:08:18 Launchpad bug 1233882 in heat "neutron resources don't have the "show" attribute as used in the example templates" [High,Fix committed] https://launchpad.net/bugs/1233882 20:08:38 If we need to, we'll fix them and do an RC2 20:08:46 sounds good 20:09:01 Any other comments or questions re the Havana release? 20:09:31 really, please everyone do test it, as we're still finding pretty significant issues.. 20:09:37 shardy, the f17 LB? 20:10:01 I have a local f19 one 20:10:11 but you wanted to go to f18? 20:10:16 asalkeld: Yeah, we should bump that to F18/19 20:10:27 I'll make a bug for that 20:10:28 asalkeld: I don't mind, anything that works and isn't EOL is OK with me 20:10:47 let's fix it in master after the branch and we can consider backporting 20:11:08 asalkeld: Can we reimplement the LB as a provider template now, ie during Icehouse? 20:11:10 has anyone tried the neutron loadbalancer? 20:11:13 seems like we probably can 20:11:48 yea, I can give that a go 20:12:20 asalkeld: cool, would be great to get rid of all the hard-coded nested templates :) 20:12:34 yip 20:13:23 https://bugs.launchpad.net/heat/+bug/1234375 20:13:26 Launchpad bug 1234375 in heat "the aws loadbalancer use Fedora 17 and needs to be upgraded to f18 or f19" [Undecided,New] 20:13:43 #link https://bugs.launchpad.net/heat/+bug/1234375 20:14:31 Ok, well great work everyone on Havana, we delivered 42 blueprints and 251 bug fixes! 20:14:48 lots of bug fixes 20:14:52 yeah, that's good 20:14:55 thanks to all, looking forward to the next push for Icehouse 20:14:57 how many bug creations? ;) 20:15:05 zaneb: ha, true.. 20:15:15 feature dev == bug creation 20:15:26 :) 20:15:32 Yeah, hence my plea for testing :) 20:15:39 #topic open discussion 20:15:49 so when is the summit sessions closed? 20:16:14 So I think they need to be wrapped up in the next couple of weeks, by mid Octover 20:16:18 October even 20:16:27 ok 20:16:31 the ides of October 20:16:41 ok, sorry I got distracted for the body of the meeting 20:16:47 I've not approved any, as I expected we'd want some discussion (at next week's meeting), and the new PTL can approve them 20:17:16 I've actually got a couple to propose too, so if anyone else has, please add them before next week: 20:17:32 #link http://summit.openstack.org/ 20:17:46 the ui lets you join stuff together or break them apart as well 20:18:01 sdake_: Cool, that will be useful 20:18:03 new ptl can expect to spend 4-5 hours preparing those talks 20:18:11 I'll be a bit quiet this week (ear infection) 20:18:28 13 proposed sessions so far 20:18:34 #info summit session agenda to be discussed at next week's meeting 20:18:43 Do you plan to have Software orchestration design meeting? 20:18:44 #action all to add remaining summit proposals 20:18:53 gokrokve: yes 20:19:12 gokrokve: yes, the software config BP related discussion definitely warrants a session 20:19:16 (IMHO) 20:19:29 Can we start some discussions before summit? 20:19:37 stevebaker: will you be quiet, or will it just seem quiet to you? 20:19:39 mailing list 20:19:40 * zaneb ducks 20:19:58 gokrokve: Yes, of course, but I wanted to ensure we have the full list of proposals before we start prioritizing too much 20:19:58 I want to call attention to the thread that therve started on the mailing list about autoscaling and load balancing. there's a workable solution being proposed but it needs some discussion 20:19:59 you need to start a etherpad gokrokve 20:20:08 gokrokve: yes, I'm hoping to write some design docs to kick off some pre summit discussion 20:20:33 stevebaker: That will be great. 20:20:42 asalkeld: +1, etherpads will help with the discussion re priorities too 20:20:44 gokrokve: are you on Murano? 20:20:58 ie details as some of the proposals are pretty sparse 20:21:12 stevebaker: Yes. 20:21:15 one thing that i would have found handy in previous summit years was a master list of etherpads in one etherpad eg summit2013novheat 20:21:18 or something similar 20:21:22 with links to each specific etherpad 20:21:34 +1 20:21:40 +1 20:21:41 We see that we have some code which actually better fits to Heat orchestration agenda then to Murano. 20:22:15 gokrokve: OK, at some point I'd like to understand why Windows software config needs a whole new API - or maybe I don't ;) 20:22:17 gokrokve: yes, that sounds really interesting and we definitely want to discuss that 20:22:34 zaneb: +1 20:22:49 stevebaker: That is the whole point. It does not require specific API. 20:22:56 sdake_: Sounds like a good idea 20:23:04 #link http://summit.openstack.org/cfp/topic/8 20:23:09 that's the heat proposal list 20:23:10 stevebaker: So we want to pass this stuff to Heat instead of doing this by ourselves. 20:23:24 shardy non-ptls get "forbidden" on that link 20:23:25 that URL gets me "Forbidden" 20:23:40 website presents different urls for ptl vs nonptl 20:23:41 gokrokve: I'm hoping that the software config proposal I'm putting together will fit well with Murano, even if you're writing your own non cloud-init agent 20:23:43 yeah that's a real bummer 20:23:50 Hmm, ok sorry, I'm not sure how to get a topic filtered link then 20:24:05 ctrl-f "Heat" 20:24:06 you have to link them one at a time from your ui 20:24:08 pretty weak :) 20:24:24 seriously, only the PTLs can see a list 20:24:35 why are the openstack web devs so user-hostile? 20:24:46 I thought you'd all see the list, only without the review option 20:24:48 sigh 20:24:53 what did we do to them? 20:24:54 stevebaker: our agent is very simple it actually listens on rabbitMQ queue and receive scripts and commands to execute. It can be anything except cloud-init. 20:25:09 zaneb: you know what you did… ;) 20:25:12 you can sort the list by project 20:25:24 one suggestion is to make a master list with links to etherpads 20:25:26 gokrokve: so are you using cloudbase-init, or did you replace it with your agent? 20:26:05 tspatzier, that's as good as it gets 20:26:06 randallburt: I may have questioned their parentage once or twice, but only after they started pulling this stuff ;) 20:26:16 shardy: we use cloud-init to do initial first boot configuration. After that cloud-init does not work on Windows as it is executed only once ant first boot. 20:26:39 lol 20:26:43 gokrokve you can use cfn-hup for that 20:26:47 shardy: So we have to use our own agent to execute commands after VM restart. 20:26:48 gokrokve: if you really wanted to align with Heat you could poll for metadata changes instead of listening to rabbitmq 20:27:27 sdake_: I think yes. We can use anything that can execute scripts. 20:27:33 hmm 20:27:34 Yeah, there seem to be significant areas of overlap - and an FAQ from users I've spoken to is Heat windows instance support 20:27:48 (which we currently don't have) 20:27:49 well lets not invent yet another whole set of agents for a new project 20:27:56 lets try to have one set of agents :) 20:28:10 sdake_: Totally agree. 20:28:18 I take it all back, go to http://summit.openstack.org/ and click on 'Topic' and it sorts the list 20:28:21 agent overhead is hard on everyone - image builders etc 20:28:32 sdake_: We created windows agent as there were no one. 20:29:06 gokrokve: so you should look at os-collect-config for fetching the metadata, then use that to trigger your own tool to do the configuration 20:29:39 end of meeting? 20:29:53 I have another BP to discuss. 20:30:01 ok 20:30:10 https://blueprints.launchpad.net/heat/+spec/stacks-reservation 20:30:14 any idea why heat doesn't show up in the Openstack-dev subscribe list? http://lists.openstack.org/cgi-bin/mailman/options/openstack-dev 20:30:21 This is for Climate. 20:31:04 We have a team who now actively works on CLimate development. They want to integrate soem CLimate capabilities to Heat. 20:31:20 Can someone review this BP and related proposal on Wiki? 20:31:22 spenceratx: because nobody added it 20:31:35 So Climate would be a separate service, making a reservation does not directly interact with any scheduler? 20:31:40 ok 20:31:51 spenceratx: but I don't have an answer to your follow-up question ;) 20:32:20 gokrokve, sounds interesting 20:32:34 zaneb: touche 20:32:46 here's that thread I mentioned earlier. could use some input at some point :) http://lists.openstack.org/pipermail/openstack-dev/2013-October/015912.html 20:33:02 #link http://lists.openstack.org/pipermail/openstack-dev/2013-October/015912.html 20:33:06 thanks :) 20:33:24 Our enginners who are working on CLimate are in UTC+3 timezone. It would be nice if we can chat in IRC or talk around 8-9am PDT 20:33:47 gokrokve: We can review the BP, we're still getting Havana finished, so there are lots of Icehouse (or potential Icehouse) BP's which are unreviewed at this point 20:33:51 what would be the atomic operation? 20:34:03 gokrokve: that blueprint is currently too high-level to act on, it really needs some climate people to break it down into some more specific feature blueprints 20:34:37 Yeah, looking at the wiki page I'm not even sure what it has to do with Heat at all 20:34:46 needs much more detail 20:35:04 If the atomic operation is reserving all the resources of a stack at once, I see a connection to Heat 20:35:05 shardy, well they could have a reservation resource 20:35:05 shardy I think they want heat to tell climate about the reservations 20:35:33 and the other resources in the template ref' the reservation 20:35:39 MikeSpreitzer: atomic means that you will reserve the whole bunch of resources (IP, VM, floating IP, volume) as a one object. At least this is my understanding. 20:36:11 I read it as Climate (Reserve this stack) -> Heat -> Climate (Reserve these individual resources) 20:36:13 Ok, well lets say that in the BP rather then "Heat stacks Reservation support for the OpenStack Reservation service" 20:36:49 So something like 'heat stack-reserve' that guarantees that a later 'heat stack-create' will succeed since you got sufficient resources? 20:37:08 i dont think you need a new api call 20:37:09 Here is more details: https://wiki.openstack.org/wiki/Heat/Reservation 20:37:21 you can just stack-create and it will reserve and create atomically vs what is done now 20:38:15 gokrokve: Yeah, I'm saying that page needs more specifics, what does Heat need, new Resource implementation, API integration with existing services, new API actions, or ..? 20:38:16 sdake: yeah, agreed 20:38:26 sdake: reservations can be created well in advance of creating the stack 20:38:30 sdake, yeah, that's another way to look at it. 20:38:41 shardy: Ok. Makes sense. 20:38:55 If a reservation is created well in advance, what happens to the physical capacity in the meantime? 20:38:56 these sound like questions for gokrokve to answer zaneb, radix, tspatzier :) 20:39:00 What will be the best to discuss the details? openstack-dev ? 20:39:12 gokrokve: +1 20:39:39 gokrokve: Are climate peeps coming to Hong Kong? 20:39:49 stevebaker: Yes. 20:40:10 Ok, shall we wrap up the meeting and continue this on the ML? 20:41:06 shardy: Yes. We will proceed with the discussion in ML for both Murano and Climate. 20:41:19 Ok then, lets do that, thanks all! 20:41:23 #endmeeting