14:01:14 #startmeeting ovirt weekly sync 14:01:14 Meeting started Wed Jun 13 14:01:14 2012 UTC. The chair is mburns. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:01:14 Useful Commands: #action #agreed #help #info #idea #link #topic. 14:01:26 #topic roll call and agenda 14:01:33 * aglitke is here for vdsm 14:01:37 * jb_netapp is here 14:01:41 * mgoldboi here 14:01:44 RobertM: yes; I had to re-run the engine-setup a second time, but it does come up 14:02:02 Agenda: 14:02:03 Status of Next Release 14:02:03 Go over details for the upcoming release day. 14:02:03 Sub-project reports (engine, vdsm, node) 14:02:03 Upcoming workshops 14:02:04 How can Red Hat OSAS (Open Source and Standards) team help with oVirt? (Dave Neary) 14:02:04 * lpeer here 14:02:05 RobertM: I'm failing on attempting to create a new nfs data domain 14:02:05 Making the Infrastructure team a sub-project 14:02:10 * rickyh here 14:02:17 rharper is here 14:02:22 * Guest3690 is here, available most Wed morn now 14:02:36 * dneary says hi 14:02:41 * garrett says hi too 14:02:50 mburns, test day? 14:02:52 rharper, I had the problem last night. I engine-cleaned then skipped that step and it completed fine 14:03:17 * lh is here 14:03:21 aglitke: sure, we can discuss the test day too 14:03:25 * RobertM is here 14:03:56 #topic Status for next release 14:04:06 mburns, I sent an email asking that testing day be added to the agenda 14:04:21 RobertM: not the initial nfs setup, but once I'm in the admin pane, adding a new data domain (nfs) ... it's just sitting there with the spinning wheel... 14:04:27 RobertM: yep, we'll discuss 14:04:56 rharper, I can't get that far. jboss wont load for me. 14:05:07 rharper, You are interrupting a meeting :) 14:05:29 #link https://bugzilla.redhat.com/show_bug.cgi?id=822145 14:05:44 aglitke: I know ... 14:06:05 * ofrenkel here 14:06:05 ok, looking at the release bugzilla above, 2 bugs still open 14:06:14 #link https://bugzilla.redhat.com/show_bug.cgi?id=824397 14:06:20 #info status: Modified 14:06:37 is this in the latest RPMS under Beta? 14:06:52 mgoldboi: do you know? ^^ 14:07:35 lpeer: ^^ 14:07:42 mburns: it's pushed in latest rpm 14:07:43 s 14:07:53 mgoldboi: ok, excellent 14:08:00 #info in latest beta rpms 14:08:00 mburns, I just installed the /beta last night and it pulled in F17 jboss packages. 14:08:22 #link https://bugzilla.redhat.com/show_bug.cgi?id=824420 14:08:29 this is ovirt-node f17 support 14:08:51 #info very close to completion on this bug, one issue still remaining, then build and post 14:09:05 mburns: though we need a re-spin on those rpms due to some bugs, i wanted to do a rebase if no extensive testing has began 14:09:45 mgoldboi: ok 14:10:10 mburns: i'll build it and send you the relevant rpms location 14:10:13 mgoldboi: anything being pulled in a rebase that isn't essential to 3.1 release? 14:10:26 or is everything relevant? 14:10:46 mburns: lots of it is relevant in terms of bug fixes 14:11:08 mburns: i think we're ok with a last rebase and then start cherry-picks 14:11:19 mgoldboi: ok 14:11:28 mgoldboi: when are the new packages going to be ready? 14:11:37 mburns: today 14:11:39 mgoldboi: we have afew network patches I'd like to push today can you please ping me before rebase? 14:11:45 #info engine will rebase to latest and rebuild 14:11:55 #info new packages to mburns before EOD 14:12:16 mburns, Are the current packages on ovirt.org ok for testing tomorrow? 14:12:19 except for node 14:12:25 #info ovirt-node will try to post something for tomorrow 14:12:27 lpeer: sure - but let's get done with it in an hour or two 14:12:40 aglitke: engine looks like there will be new packages coming today 14:12:41 aglitke: no, they are not. 14:12:42 I was hoping to be installed today so I don't have to waste the whole day tomorrow with install issues. 14:12:48 mgoldboi: will do 14:13:13 danken: ping -- what's the status of vdsm? 14:13:25 are there new packages coming for the test day? 14:13:56 current version is vdsm-4.9.6-2.gite952471.fc17 14:14:21 we are supposed to be cutting to 4.10.0 for the release. 14:14:32 Not sure if that is planned for now or later. 14:14:47 mburns, wrt other packages, need to verify on sdk/cli versions. 14:14:56 fsimonce: already built 4.10.0 14:15:07 and passed it to mglodboi 14:15:33 #info vdsm 4.10.0 built and will be posted today to beta area 14:15:36 itamar1: ok 14:15:48 itamar1: current sdk version is ovirt-engine-sdk-3.1.0.1-1alpha.fc17 14:16:02 cli is the same, just s/sdk/cli 14:16:11 please send out a note as soon as all of the rpms have been posted. 14:16:21 aglitke: will do 14:16:32 #action mburns to update when packages are posted to beta area 14:16:48 itamar1: i don't see ovirt-guest-agent rpms in beta 14:16:55 are there versions of that coming too? 14:17:09 mburns, on sdk/cli - sdk should be 0.1, but cli should be 0.2 14:17:19 y. 14:17:21 bazulay --^ on guest agent rpms 14:17:34 itamar1: who is responsible for getting the rpms built? 14:18:00 guest agent rpms - bazulay/gal i believe 14:18:17 sdk can be 0.2 as well, but cli 0.2 can work with 0.1 sdk only 14:18:56 mgoldboi: can you coordinate all the rpms for beta repo? engine, vdsm, cli, sdk, guest-agent 14:19:04 basically everything but node... 14:19:13 mpastern - didn't understand - if cli should be 0.2, and can only use 0.1 sdk, then how can sdk be 0.2? 14:19:16 mburns: sure 14:19:25 mgoldboi: thanks 14:19:42 ok, to summarize: 14:19:52 mburns: please add ai on me 14:19:56 engine will rebuild today (within next couple hours) 14:20:06 itamar1: sdk 0.2 is ready, just 0.2 cli yet can work with it 14:20:07 vdsm rebuilt already 14:20:19 can't 14:20:30 need 0.1 sdk and 0.2 cli rpms 14:20:46 mpastern - then it can't be used for the release, right? 14:20:59 #action mgoldboi to coordinate rpms for engine, clie, sdk, vdsm, guest-agent and send details of locations to mburns 14:21:12 #action to post rpms to beta and send announcement 14:21:28 itamar1: i don't mind 0.2 be available as well in release 14:21:49 mpastern - i don't understand - it will break the cli, right? 14:22:08 itamar1: right, but sdk is an component by itself 14:22:24 hello 14:22:33 if we want things to work, we need to ship 0.2 for cli and 0.1 for sdk 14:22:49 mpastern, but you can't publish 0.2 of the sdk if it breaks another component without fixing that component first 14:22:50 once cli works with newer sdk, we can ship newer sdk 14:24:11 ok, so i think that covers status of the next release as well as subproject status... 14:24:29 #topic test day 14:25:10 itamar1: true, this why i limited cli to 0.1 sdk (sdk:0.1+cli:0.2), cli won't work with 0.2 sdk, it will ask for <= 0.1, this why we can release both versions of sdk 0.2 as official and 0.1 for cli 14:25:52 aglitke: RobertM: what did you want to discuss w.r.t. the test day? 14:26:15 Is there any central coordination of test activity? 14:26:36 ie. a presence in this channel? Who is participating? 14:26:48 My team plans to focus entirely on testing tomorrow. 14:26:51 Who is the point person? 14:27:06 mgoldboi: who is the point person for this? 14:27:11 Some of us will test engine setups and some will test vdsm by itself. 14:27:29 my expectation is that every project will have people here 14:27:44 assuming we get an ovirt-node out, i'll be here for sure 14:27:45 mburns: currently none, we need someone to consolidate it, can't do it this time - sorry 14:27:58 any volunteers for coordinating? 14:28:24 I cannot do it for the whole project. Definitely plan to participate though. 14:29:04 * RobertM same here plan to participate 14:29:15 The goal of the irc presence is to get quick attention on bugs found tomorrow, right 14:29:28 ykaul: anyone on your team able to help coordinate the test day? 14:29:29 so patches get reviewed and committed quickly 14:29:35 bugs get triaged 14:29:36 mburns: no, but we'll be on IRC. 14:29:46 aglitke: both that and for helping debug issues on the spot 14:29:50 yep. 14:30:17 what else do we need? 14:30:35 ideally, we'd have a wiki page similar to we had for the 3.0 release 14:30:47 where we list out test cases and track bugs that have come up 14:31:21 right, and that would need a coordinator to maintain the bug list and details. 14:31:43 mburns: no harm in re-using the same wiki page, btw. Still applicable, I believe. Don't mind going over it and refreshing to fit 3.1 if changes are required. 14:32:05 given lack of volunteers, we'll have to have a rep from each sub-project to coordinate their areas 14:32:10 ykaul: that would be a huge help 14:32:12 ykaul: thanks 14:32:14 also the coordinator would act as a traffic cop to push out some of the noise from known issues and new issues. 14:32:47 * mburns thinks he heard sgordon volunteer... 14:33:02 i had a feeling i might be volunteering for something in my absence 14:33:24 sgordon: looking for someone to play coordinator for test day tomorrow 14:33:31 sgordon: It's the old "take one step back" trick 14:33:54 dare i ask what is involved? :P 14:34:12 i think in the absence of a true coordinator, we can more or less do community coordination 14:35:02 with a rep from each team helping with their specific area 14:35:09 any objections to that? 14:35:22 mburns: +1 14:35:38 #info community coordination for test day 14:35:50 #info each sub-project will have someone helping 14:36:00 #info ykaul will do cleanup of test day page 14:36:12 #info all issues should be track on the test day wiki 14:36:21 link? 14:36:57 #link http://ovirt.org/wiki/Testing/OvirtTestDay 14:37:15 cool. I think that covers this topic 14:37:18 ok, need to move on to the next topic 14:37:29 #topic Workshops 14:37:33 mburns: that's the old test day wiki. I'll have a brand new, polished page (ok, copy-paste of the old one minus the old bugs and so on). 14:37:49 ykaul: ok, can you post the link to the new page then? 14:38:11 lh: i know you have a bunch of questions in regards to the LinuxCon Japan workshop 14:38:12 * ykaul pathetically can't find yet how to create a brand new page. Working on it. 14:38:30 mburns, yes, these can be handled here or on list as suits the team 14:38:37 can we do that over email first, then discuss results in next week's meeting? 14:38:54 mburns, sounds reasonable. 14:38:56 14[[07Testing/OvirtTestDay3.114]]4 !N10 02http://www.ovirt.org/w/index.php?oldid=3588&rcid=3678 5* 03Ykaul 5* (+13) 10Created page with "3.1 test day." 14:39:06 mburns: http://www.ovirt.org/wiki/Testing/OvirtTestDay3.1 14:39:15 #info LinuxCon Japan workshop help on Friday 8-June 14:39:19 May I ask, what is the frequency of workshops as we currently plan them? 14:39:23 mburns, please note for the minutes that we'll need everyone's help preparing the agenda for linuxcon North America 14:39:26 And how much work goes into each one? 14:39:43 #info post-mortem to be done next week for LC Japan 14:39:52 dneary, we're looking at approximately 1 per quarter, thus far scheduled at linuxcons and board member hqs 14:40:07 dneary, http://www.ovirt.org/wiki/OVirt_Global_Workshops 14:40:08 #info will need input from everyone on agenda for LC North America in August 14:40:37 dneary: LinuxCon is roughly one per quarter in a different geography each quarter 14:40:42 dneary, re: work for each one, i'd say it takes about 15 hours prep from my side, plus the team has to construct an agenda and then teach the courses, prep slides 14:40:53 we have occasional other ones outside of LinuxCon 14:40:57 a couple of those a year 14:41:44 lh: anything else for Workshops? 14:41:45 lh, OK - not a major overhead, then 14:42:01 mburns, not at this time, thanks 14:42:07 ok, then moving on 14:42:17 #topic How can OSAS team help? 14:42:21 dneary: you're up 14:42:27 Hi everyone! 14:42:35 dneary, not thus far, but that could change based on not having the Linux Foundation doing logistics legwork when we meet at board member hqs 14:42:42 #info OSAS == Open Source and Standards 14:43:10 I should start by saying that you may already know my OSAS colleagues Carl Trieloff, Karsten Wade, Jason Brooks and Leslie Hawthorn, all of whom have been doing stuff for oVirt 14:43:34 And, as mburns just said, OSAS is the Open Source and Standards group in Red Hat. 14:43:56 We're basically the A-Team for community stuff 14:44:06 Except our theme music isn't as cool 14:44:13 Anyway... 14:44:23 * RobertM lol 14:45:00 For the past couple of weeks, I have been talking to some members of the oVirt team trying to figure out whatw e might be able to do to help the project accelerate adoption and contributions to the project 14:45:20 I just sent an email on the topic to users@ 14:45:52 And I wanted to take this opportunity to introduce myself, present the team and our goal, and get feedback from the oVirt community 14:46:37 14[[07Testing/OvirtTestDay3.114]]4 !10 02http://www.ovirt.org/w/index.php?diff=3589&oldid=3588&rcid=3679 5* 03Ykaul 5* (+8406) 10 14:46:47 dneary: thanks 14:46:56 The first thing I have identified is that it's not clear from the ovirt web-page who exactly the project is aimed at 14:47:48 And I'd really appreciate help figuring out what the natural audience for the project is (and infer a target audience from that) so that we can tailor the website and outreach efforts to that audience. Does that sound reasonable? 14:48:59 dneary: i think the target is really for people who want to manage virtual machines across a number of hosts 14:49:22 so really more toward companies/enterprises rather than single individual users 14:49:35 14[[07Testing/OvirtTestDay3.114]]4 !10 02http://www.ovirt.org/w/index.php?diff=3590&oldid=3589&rcid=3680 5* 03Ykaul 5* (-182) 10 14:49:44 mburns, I have noticed that if you want to run, say, 2 or 3 VMs on one host that it's not really appropriate for that at the moment 14:49:59 dneary, Sounds reasonable to me. I am at least part of the natural audience an IT admin who wants to do as much as possible under opensource but still looking for a way to manage my server network. 14:50:08 right, the overhead is too much for a small deployment like that 14:50:27 dneary, arguably it will never be - the best features only kick in when you have multiple hosts 14:50:28 mburns, RobertM: If that were the goal, what would we suggest to people? 14:50:49 probably something like virt-manager or boxes 14:51:13 Single host virt-manager on the KVM side. 14:51:45 dneary: in relation to that it might also be good to set up the wiki and more guidelines 14:51:51 I was thinking that the natural audience for us would be something like the sysadmin who wants to set up a test lab, a feasibility study with 1 or 2 nodes and running a mgmt app from, say, a laptop 14:52:07 for example, the current naming scheme doesn't conform to the mediawiki standard, making the search rather useless 14:52:27 And from there, deploy into production once he's happy it's a good solution (either deploying oVirt or looking for a product based on it so he can get support) 14:52:44 ewoud: jbrooks and I are planning a set of "wiki gardening" sessions, we'll try to organize and publicize so others can join in the fun 14:52:54 There is no management app in ovirt. There is the engine (Master) and nodes. All management is done inside a website. A real plus in my box. 14:52:56 a service-level agreement, I mean (since of course he can get support for oVirt too) 14:52:58 (we'll also wait until the channel is quiet or shut the bot up during all the edits) 14:53:01 *book 14:53:02 quaid: good, yesterday I already added many features to http://ovirt.org/wiki/Category:Feature 14:53:10 ewoud, That is definitely a follow-on goal 14:54:09 RobertM, Yes, when I say "management app" I'm referring to the engine 14:54:16 In the context of oVirt 14:54:17 dneary: that's a pretty good standard use case 14:54:55 mburns, OK - that helps, since we can set a "minimum recommended hardware" standard for documentation and "getting started" tutorials 14:55:13 you'll find the odd individual too that will set it up and use it on their own personal machines (like me) 14:55:44 Anyway - I really just wanted to say hi, and let you know that you'll be hearing more from me and other colleagues including our new team-member Garrett Lesage (who started last week) in weeks and months to come 14:55:47 * RobertM Or buy a pair :) 14:56:03 mburns, We can of course document that, as an edge case 14:56:09 * garrett waves 14:56:12 ;-) 14:56:23 dneary: excellent, thanks for the introduction 14:56:25 But we can assume, for example, that the engine and VMs are on different hosts 14:56:26 garrett: welcome 14:56:40 howdy! (: 14:56:53 dneary, http://blog.jebpages.com/archives/how-to-get-up-and-running-with-ovirt/ is a pretty good outline of a signal box howto. 14:56:56 dneary: well, there is an all-in-one feature that let's you run both engine and vms on the same host 14:57:02 RobertM, Thanks! 14:57:15 ok, moving on to the final topic 14:57:31 #topic infra team as a sub-project 14:57:36 RobertM, I wish I knew the guy who wrote that article ;-) 14:58:14 background -- the infra team has been somewhat disorganized so far 14:58:25 dneary, He is in IRC not sure if he is online right now. Unless you are being sarcastic 14:58:47 with a few people that have access to various pieces of the infrastructure 14:58:53 RobertM, jbrooks is one of my colleagues I mentioned earlier - thus the ;-) 14:59:00 it would be good to standardize this a bit more 14:59:18 to help with the on-boarding of new volunteers (like RobertM ) 14:59:19 also is there a process to bring new people on? 14:59:49 and make sure that people know who to talk to about various problems 14:59:56 14[[07Testing/OvirtTestDay3.114]]4 !10 02http://www.ovirt.org/w/index.php?diff=3591&oldid=3590&rcid=3681 5* 03Ykaul 5* (+80) 10 15:00:33 it would also be good to start moving some of the configuration into backed up repositories as well 15:00:45 The difficulty I've seen in the past in infra teams is that there is a mix of privileged and unprivileged information in servers 15:01:00 quaid: you've been sort of the de facto lead here, any thoughts? 15:01:10 So unless you're starting (say) a VM per service, it gets tricky to give ppl admin access to just one service 15:01:24 dneary: agreed 15:01:41 but right now, there are just a few (very few) people that can fix problems that arise 15:02:10 What are the gaps in infra now? Are we running old versions of stuff, integration isn't perfect between different services, are we looking to do SSO across services, etc? 15:02:48 we're running some stuff on small EC2 instances 15:02:54 Or is it new services we'd like to add? 15:02:59 we have hacked together backups 15:03:02 Ah. Performance & price 15:03:05 (where we have backups at all) 15:03:13 OK 15:03:15 14[[07Testing/OvirtTestDay3.114]]4 !10 02http://www.ovirt.org/w/index.php?diff=3592&oldid=3591&rcid=3682 5* 03Ykaul 5* (+87) 10 15:03:19 we need a way to onboard new people 15:03:30 integration between services is lacking 15:03:45 some new services that we'd like to add, but no one has bandwidth to implement 15:03:54 I'm hoping quaid hasn't gone afk 15:04:01 no one with access has bandwidth to implement 15:04:15 * RobertM wonders if it would be better to try and get a few servers and run a RHEV cluser it might be cheaper then AWS 15:04:20 What kind of entry barrier do we want to put up for new sysadmins? 15:04:41 dneary: yes, that's another question 15:04:46 Infrastructure can quickly get very messy if ppl start adding services higgledy piggledy with no-one guiding a holistic view 15:05:24 dneary: that was really the point, we need to make infra a real team rather than just a hodgepodge of people... 15:05:34 (things like back-ups for new services, integrating services into SSO, migrating data across upgrades, maintaining an identitiy service, etc) 15:05:40 mburns, OK 15:05:52 mburns, Looks ilke we need quaid to discuss the topic 15:06:02 yes, probably best to take this offline 15:06:19 RobertM, I bet it would 15:06:21 RobertM, we are looking at some additional physical servers. AWS is expensive and worse, provides very bad performance 15:06:35 i'll put together an email to infra@ and we can start there 15:06:41 But then we have the Q of the physical & virtual access to the servers to consider. 15:07:01 Personally I don't mind - the GNOME Foundation has servers scattered around the planet in half a dozen colos 15:07:44 i think the first thing to do is nominate a infra manager 15:07:53 and let them start the coordination 15:08:01 dneary, I guess it comes down to budget, requirements, and team member location. 15:08:08 * mburns immediately steps back so he doesn't get nominated... 15:08:35 * RobertM Dang now who do I suggest since mburns is running for the hills :) 15:08:48 mburns, If you're going to run it like a team, then yes, you need an active maintainer 15:09:03 RobertM, Let's take it to the list (which list?) 15:09:06 RobertM: i'd need to decline, i have too many other things on my plate to take on that type of role 15:09:18 infra@ would be where i'd start 15:09:25 mburns, Ah, of course 15:09:40 we can escalate to board@ if we need board input 15:10:27 ok, any other comments? 15:10:31 or other topics? 15:10:42 I'm afraid I have to go in a sec. 15:11:28 ok, starting the countdown 15:11:31 going once... 15:12:02 hmm, sort of missed the infra as a sub-project discussion 15:12:07 hi i'm starting to try ovirt, i'd like to know if the ovirt node image deployment is needed? 15:12:13 mburns, To confirm everyone things making infra as subproject is a good idea 15:12:20 but at fosdem there was a presentation about how they did it, maybe look at how they did it? 15:12:24 things=thinks 15:12:31 s/they/wikimedia 15:12:44 ewoud: not a bad idea 15:14:02 http://blog.wikimedia.org/2011/09/19/ever-wondered-how-the-wikimedia-servers-are-configured/ might be interesting in that regard 15:14:11 mburns, With detail to be convered on the infra list? 15:14:51 RobertM: yes, we'll take this to infra and make decisions from there 15:15:02 Thank you all for the great welcome! I have to go. Goodbye! 15:15:12 ok, last call for additional topics? 15:15:49 going twice.. 15:16:00 gone. 15:16:03 thanks all! 15:16:06 #endmeeting