21:00:26 #startmeeting 21:00:27 Meeting started Tue Jul 12 21:00:26 2011 UTC. The chair is jaypipes. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:28 Useful Commands: #action #agreed #help #info #idea #link #topic. 21:00:29 \o 21:00:51 Thierry is away today; soren and I will fill in this week... 21:00:53 o/ 21:00:55 o/ 21:00:55 o/ 21:01:07 * soren adopts a French accent 21:01:14 * jaypipes dons beret 21:01:21 oui? 21:01:21 You've lost me. 21:01:37 \o 21:01:57 OK, let's start, eh? 21:02:03 jaypipes will henceforth be known as "Don Beret" 21:02:09 heh 21:02:18 #topic Swift status 21:02:26 notmyname: any updates? 21:02:33 I don't think there are any updates 21:02:43 point of order: Is there a link to the agenda? 21:02:47 k. anything you need assistance on from anyone? 21:02:51 Vek: one sec 21:03:14 http://wiki.openstack.org/Meetings/TeamMeeting 21:03:28 dang, beat me to it... 21:03:31 we saw some packaging updates go through this morning for swift. I assume those a work in progress? 21:03:36 soren: ^? 21:03:38 Looks quite boiler plate-y. 21:04:14 notmyname: Sort of? 21:04:23 soren: some details, please? :) 21:04:26 I mean... EVerything is work in progress :) 21:04:57 I think there were some concerns about some of the init scripts. I don't have the details, but we can get those to you 21:05:02 I expect to start merging the stuff you've done tomorrow. 21:05:12 o/ 21:05:15 ..but the stuff that landed today isn't meant to be half-done if that's what you're asking. 21:05:30 soren: ok. then we definitely need to comment ;-) 21:05:57 notmyname, soren: #action get together on Swift packaging needs 21:06:25 notmyname, soren: can you two handle that off list? or is it something we should go over here? 21:06:33 not here 21:06:36 k. 21:06:38 I have no idea. 21:06:45 notmyname: any other stuffs? 21:06:53 (seeing as I seem to be the one who'll be told thing, not telling things) 21:07:08 jaypipes: I don't think so 21:07:21 any questions for notmyname before we move on? 21:07:48 #action notmyname and soren get together on Swift packaging needs 21:08:08 ok, moving on... 21:08:12 #topic Glance status 21:08:16 https://launchpad.net/glance/+milestone/diablo-3 21:08:52 Mostly been handling bugs this week and last couple days of last week. Big one almost finished (just finishing tests on it): https://bugs.launchpad.net/bugs/771849 21:08:53 Launchpad bug 771849 in glance "Port not parsed correctly in Swift URI" [High,In progress] 21:09:50 have a small blocker on getting https://bugs.launchpad.net/bugs/713154 done: waiting to set up an AWS testing account we can put creds on the Jenkins box for... 21:09:51 Launchpad bug 713154 in glance "S3 Backend doesn't support POST either." [Medium,In progress] 21:10:10 Have to get back on track with the Keystone integration as well... 21:10:28 Still hopeful to hit D3 for that, with shared image groups in D4 21:10:48 #action Jay to update D3/D4 blueprints 21:11:55 The recent change in nova to support multiple glance servers threw a wrench in a bunch of people's setups. We should try to avoid such situations better in the future... the config files changed (from glance_host/port to glance_api_servers FLAG and things weren't upgrade cleanly) 21:11:58 jaypipes - we are looking at Keystone integation, too 21:12:28 glenc: resource-wise, do you have a specific allotment for someone to work on that? 21:12:46 glenc: I know you, westmaas and maybe pvo needed to chat about that? 21:12:48 we could - we were hoping to start it next sprint (Monday) 21:13:00 glenc: k, I'll offline it with you. 21:13:02 we should sync - don't want to duplicate work 21:13:05 yup. 21:13:11 that's about it from me. any questions on Glance stuff? 21:13:30 in related news, JordanRinke has Glance working on Windows now.. w00t. 21:13:50 nice 21:14:05 ok, moving on... 21:14:06 jaypipes: is there a walk-through anywhere for loading in images into Glance? I've been looking, haven't spotted one. 21:14:21 (sorry) 21:14:45 heckj: there's a variety of ways to do so. Here is a good starting point:http://glance.openstack.org/glance.html#the-add-command 21:15:12 jaypipes: thanks 21:15:21 heckj: if you're using the eucatools, you can check out the glance-upload tool, too. 21:15:42 ok, moving on... 21:15:49 #topic Nova status 21:15:53 vishy: you're on. 21:18:08 *crickets* 21:18:09 OK, ttx wanted me to ask mtaylor and soren about your blueprints targeted for D3 Nova: https://launchpad.net/nova/+milestone/diablo-3 21:18:37 soren: https://blueprints.launchpad.net/nova/+spec/ec2-id-compatibilty I assume is held up on the ML discussion on EC2 UUIDs? 21:18:48 jaypipes: Sounds like a fair summary. 21:19:01 jaypipes: I also have a ton of packaging things to do and some holiday and OSCON and and and.. 21:19:15 mtaylor: your 2 testing-related bluprints for Nova's D3? Any updates? 21:19:16 If someone else wants it, feel free to jump on it. 21:19:26 wow 21:19:32 sorry got disconnected :) 21:19:45 soren: I think based on the discussion on the ML that one might need to get pushed back... 21:19:50 vishy: no worries, mate :) 21:20:03 vishy: have at it. 21:21:12 i missed some scrollback unfortunatley 21:21:18 *ely 21:21:28 vishy: just asked soren and mtaylor about their Not Started blueprints for D2 21:21:31 sorry, D3 21:21:45 vishy: you also have a Not Started blueprint for separating services... 21:22:01 I'm thinking of cancelling that one 21:22:17 our next sprint is to spend some time updating and fixing the volume service 21:22:28 which may provide some results along those lines 21:22:42 but it is probably best to just cancel that blueprint and make a new one 21:22:44 vishy: any particular blueprints or bugs that you need extra attention on? any particular merge proposals that really need reviews? 21:22:50 yes 21:23:06 I would love some feedback on the ha-networking branch that I proposed 21:23:32 also, since lunr has moved to the driver level, we are going to need some more focus on the volume service 21:23:41 there will be a few new blueprints around that showing up 21:23:46 vishy: I'll send out a "important stuff" mailing list post like ttx usually does. 21:23:51 and we might need some help 21:24:21 i'm a little concerned about full keystone integration making it in 21:24:36 vishy: how does the lunr thing work now? have we given up on the idea of elastic block storage at the API level? 21:24:46 the main sticking point that no one has tackled is making ec2 api work properly 21:25:03 jaypipes: volume already does ebs 21:25:06 vishy: keystone and EC2? 21:25:25 they have just decided to implement lunr as a driver and keep the existing nova code in place 21:25:40 vishy: right, but it's a subset of the compute API now, correct? no plans to sepaarate that out as its own service? 21:25:55 jaypipes: If you would have read the email, I stated that we were scaling back the focus of lunr to the driver level 21:25:56 jaypipes: it depends on how you define service 21:26:21 jaypipes: there is already a nova-volume service 21:26:34 whether it ever graduates into a separate project is unknown 21:26:40 creiht: sure, but I didn't understand that, really, since nova already contained driver-level code for volumes... 21:27:50 jaypipes: oh really? :P 21:28:16 creiht: yes. 21:28:27 I guess that will work then, my job is done 21:28:32 creiht: there are drivers for AoE, iSCSi, HP LeftHand, etc 21:28:45 jaypipes: on the keystone ec2 stuff, if we depend on keystone completely for users and tenants, we don't really have a good place for generating/storing/checking ec2 access and secret keys 21:28:51 creiht: are you saying that you guys will be adding additional drivers only? 21:29:02 jaypipes: that is correct 21:29:07 creiht: k, thx 21:29:27 jaypipes: this either needs to be a keystone plugin or move into an ec2 only compatibility layer 21:29:36 vishy: k, understood. do we have a gameplan for that? /me crosses fingers... 21:29:47 on volume-service stuff: there are multiple things that might be required... starting from scheduling and reporting capabilities a-la compute 21:29:57 since nova will ultimately lose its concept of user and project 21:30:00 jaypipes: and helping push nova volume in feature and other areas where it lacks 21:30:25 vladimir3p: yeah, I'm working on blueprints right now for some of those areas 21:30:31 creiht: coolio. good to know. I re-read the ML post. thx. 21:30:34 vladimir3p: exactly. nova-volume got a little bit abandoned over the last few weeks and it needs to be brought up to speed 21:30:39 Folks, we have 2 blueprints on our plate: VSA & adminVM. Both are "low priority", but quite important for us. We plan to submit VSA up to the end of this week, but AdminVM is stuck for ~1.5 months without any valuable feedback. The VSA blueprint touches volume service & scheduling and we could try to integrate them better... 21:31:10 our VSA code has multiple changes related to scheduling, etc 21:31:26 vladimir3p: I thought there was some useful feedback on that proposal 21:31:31 will be nice to find the common ground 21:31:51 jaypipes: thank you for your undivided attention 21:31:57 i think the majority of the feedback was to simplify it to arbitrary network plugging + metadata 21:31:57 frankly, have not seen any feedback on VSA yet ... 21:32:08 creiht: thank you for your undivided sarcasm. ;) 21:32:08 maybe I'm thinking of a different proposal 21:32:48 vishy, vladimir3p: shall I add the AdminVM to thierry's post of important things for reviewers? sounds like I should? 21:33:00 jaypipes: yes please 21:33:05 k, will do. 21:33:17 thanks 21:33:21 no pros. 21:33:27 no prob even... 21:33:44 alright, vishy, any other stuff to add to the important list? 21:34:12 i think that is good for now 21:34:26 any more questions for vishy? 21:35:01 alrighty... thx everyone. same bat-time, same bat-channel next week. 21:35:04 #endmeeting