14:02:24 #startmeeting oVirt Infra 14:02:24 Meeting started Mon Oct 14 14:02:24 2013 UTC. The chair is knesenko. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:02:24 Useful Commands: #action #agreed #help #info #idea #link #topic. 14:02:33 #chair ewoud eedri_ obasan 14:02:33 Current chairs: eedri_ ewoud knesenko obasan 14:02:41 Rydekull: here ? 14:02:43 * orc_orc is here 14:02:56 #chair orc_orc 14:02:56 Current chairs: eedri_ ewoud knesenko obasan orc_orc 14:03:13 orc_orc: Hello 14:03:24 good day ;) 14:03:31 orc_orc: would you like to introduce yourself ? 14:03:31 orc_orc, hello 14:04:02 I am Russ Herrold ... was at the meeding last week, and getting a handle on the infra suport of the CI system presently, with an eye to replicating it 14:04:47 orc_orc: welcome aboard 14:04:51 thannk you 14:05:22 orc_orc: on which areas do you prefer to work ? system administration ? coding ? 14:05:22 orc_orc, are you considering joining the infra team as a formal member? 14:05:53 eedri: I am happy to do so ... I run such in other contexts 14:06:43 I do not quite understand the approach to reporting and solving infra bugs as it does not seem an issue tracker is systematically used, other than a mailing list 14:07:28 orc_orc: you are right ... we use mailing listing more than the issue tracker 14:07:37 we should switch to it 14:07:40 ok lets start 14:07:46 #topic Hosting 14:07:50 ok ... 14:07:57 knesenko: I saw your migration plan 14:08:00 so I sent the migration plan to the infra list 14:08:19 so you are more then welcomed to review it 14:08:23 ewoud: how it looks like ? 14:08:29 http://lists.ovirt.org/pipermail/infra/2013-October/004123.html 14:08:40 knesenko: I was wondering what you meant by the mac address mapping 14:08:50 ewoud: ah ... 14:09:10 ok .. so now we have static mapping between the VMs mac. 14:09:30 what I mean is: vm01 is running on rackspace01 and it works ok 14:09:48 If we will try to migrate to another host, VM will lost connectivity 14:10:07 because rackspace guys did some mapping on their side 14:10:29 So I asked them to make it dynamic, so we will be able to migrate VMs between hosts 14:10:45 knesenko: it would be helpful to document the network setup there because this is new to me 14:10:55 ewoud: yes ... right 14:11:16 #action knesenko document network setup for hosts at rackspace 14:11:21 knesenko: it would seem an 'assignment pool' of MAC addresses is needed betwrrn a given cluster of dom0's to manage, so that they could self manage this at layer 2 ? 14:11:31 ewoud: that's what I wrote ... you are faster than me 14:11:32 knesenko: other than that I think it's good 14:11:53 knesenko: can we use the ovirt mac address pool feature for this? 14:12:16 orc_orc: maybe ... I opened a ticket. lets wait for their reply 14:12:33 ewoud: I am not aware of that feature :) 14:12:39 ewoud: what is that ? 14:12:57 knesenko: I changed it for alterway02 because we had a conflict because ovirt always defaults to the same 14:13:43 knesenko: but ovirt has a pool where it picks a mac address from when you create a new VM 14:13:52 ewoud: ah ... you mean the way engine generates the MAC address pool ? 14:13:55 knesenko, ewoud is talking about MacPoolRanges field in vdc_options table 14:14:05 knesenko: yes 14:14:08 ewoud: ah ok ...its not relevant in our case 14:14:21 knesenko: how so? 14:14:26 ewoud: the problem is in their network configuration 14:15:09 according to their netowrk VM with mac A will be able to connect to the internet only when running on host B 14:15:16 that's their policy 14:15:49 knesenko: ah 14:16:08 and we want setup with migration 14:16:45 so ... they need to solve and define that VM with MAC A should be able to run on rackspace* 14:16:49 all our servers 14:16:55 ok anything else on hosting ? 14:17:12 good 14:17:21 #topic puppet and foreman 14:17:25 ewoud: news ? 14:17:30 dcaro is not here today 14:17:59 knesenko: no, had no time to work on it last week 14:18:35 knesenko, yes 14:18:45 eedri: yes please 14:19:01 knesenko, we have a few open issues for puppet, i wonder if i should add it as tickets to trac 14:19:29 eedri: yes ... I would like to move all tasks to track 14:19:56 guys, if you see email with some request - please reply to them to open a ticket . 14:20:01 we should start using tickets 14:20:34 knesenko, ok, we have some issues, i will open tracs for them 14:20:44 eedri: good 14:21:03 orc_orc: are you familiar with puppet ? 14:21:36 #action eedri to open trac tickets on open puppet tasks 14:22:18 knesenko: yes -- I use it locally for some projects 14:22:32 I was looking for the trac link -- the web pages seem to point to bugzilla as the issue tracker ? 14:22:54 orc_orc: send me the link please 14:22:55 we should update the trac pages 14:23:36 both: http://www.ovirt.org/Community and http://www.ovirt.org/Communication seem to 14:23:59 https://fedorahosted.org/ovirt/report 14:24:02 orc_orc, ^ 14:24:05 http://www.ovirt.org/Infrastructure_team_meetings 14:24:06 orc_orc: ovirt uses bugzilla where ovirt infra uses trac so that may be a bit confusing 14:24:30 oops -- that page needs the link -- shall I add it? 14:26:18 orc_orc: please 14:26:29 anything else ? 14:27:27 #topic Jenkins 14:27:28 knesenko, yea 14:27:32 knesenko, on jenkins i mean 14:27:35 eedri: :) 14:27:39 shoot 14:27:57 knesenko, there were some requests by community for fedora20 slave and ubunto slave 14:28:19 knesenko, i think we can reinstall one of the f18 slaves as f20 14:28:28 eedri: ok .. lets open a tciket for that also 14:28:30 knesenko, since most focus now is on f19, right 14:28:34 eedri: ok ... 14:28:42 ewoud, can foreman install both? 14:28:49 eedri: make no sense to me to do it before the migration 14:29:14 eedri: yes, but we may need to add the preseed template 14:29:26 orc_orc: would you like to handle puppet tasks eedri will open ? 14:29:46 eedri: https://github.com/theforeman/community-templates/tree/master/preseed should be a good starting point 14:31:15 ewoud, are these ready made iso files ? 14:31:26 ewoud, or provision temaplres 14:31:33 eedri: provision templates 14:31:59 ewoud, ok, 14:32:17 #info info on install ubunto in foreman: https://github.com/theforeman/community-templates/tree/master/preseed 14:32:43 ewoud, what about f20? 14:33:23 knesenko: I am happy to look at them and try -- my account name in FAS is: herrold 14:33:39 orc_orc, you just need user on gerrit.ovirt.orgt 14:33:43 orc_orc: great ! 14:33:51 orc_orc, then you can push patches to any ovirt project 14:34:05 eedri: that should be a matter of adding the Operating System in foreman, link the templates and install 14:34:36 ewoud, ok 14:34:44 ewoud, so i guess we'll need to add it as ticket as well 14:35:07 eedri: yes 14:36:14 #action open ticket on adding support for installing f20 and ubunto in foreman.ovirt.org 14:36:34 anything else here ? 14:37:17 knesenko, yes 14:37:32 knesenko, jenkins LTS has newer version with multiple bug fixed, needs to be upgraded 14:37:44 obasan: ^^ ? :) 14:37:46 knesenko, also, lots of plugins versions are also outagred 14:38:01 eedri, knesenko I will do it 14:38:43 obasan: good 14:38:58 #action obasan to upgrade jenkins.ovirt.org to latest LTS and review plugins upgrades 14:39:11 knesenko, also, i'm currently testing adding a new bare-metal slave to jenkins 14:39:26 knesenko, you can see it as "remote_slave_jnlp_dell01" 14:39:58 eedri: good ! 14:40:11 knesenko, so feel free to try and run jobs on it and play with it 14:40:39 where is Rydekull ? 14:40:42 #info new bare-metal minidell added as jenkins slave to testing (will be able to run VMs) 14:41:36 I will try to send him email 14:42:05 This week I"ll try to finish with rackspace migration 14:42:09 I hope so 14:42:19 seems like we are done here. right ? 14:42:50 knesenko, maybe it's worth review tickets? 14:42:55 +1 14:43:02 knesenko, cause i'm not sure anyone looks at them outside of this meeting 14:43:13 eedri: I need to go now ... If you can take a lead on it it will be great 14:43:15 eedri: can you ? 14:43:34 knesenko, i'm not fully familiar with the bot commands 14:43:53 at the end of the meeting just run \#endmeeting 14:43:55 :) 14:44:00 eedri: that's all 14:44:10 knesenko, ok 14:44:14 eedri: thanks 14:44:17 buy guys 14:44:33 ok, last things on jenkins 14:44:42 i saw new vdsm functional tests job 14:44:52 eedri: 16:02:24 <@ovirtbot> Useful Commands: #action #agreed #help #info #idea #link #topic. 14:45:01 i think apuiemo is managing them 14:45:25 * eedri doesn't remember the exact nic 14:46:15 there was something with the maintainer of that 14:46:19 wasn't it passed on? 14:46:29 i think we should appoint an engine and vdsm owner for jenkins upstream 14:46:59 someone who will be in charge of monitoring the jenkins status, cause right now some jobs might stay broken for some time 14:47:11 like it was, without anyone looking into it 14:47:29 eedri: I do like the idea and jenkins can easily always CC someone when a build breaks 14:47:36 and last time i proposed to add infra@ovirt.org as email recepient on failing jobs, it was rejected 14:48:03 eedri: I think because infra@ was getting spammed and infra may not be able to fix the jobs 14:48:16 ewoud, yes, i agree 14:48:17 if it's code for example 14:48:30 ewoud, there are a few ways to handle it, we can tell jenkins to send only 1 email per failure for e.g 14:48:47 ewoud, so it won't email on reoccuring failures 14:49:02 ewoud, or maybe a job that will email unstable jobs daily/weekly? 14:49:04 to infra 14:49:18 eedri: I like the daily/weekly idea 14:49:38 ewoud, or we can spend time in the meeting to review them? 14:49:55 eedri: I think a weekly mail before the meeting may be helpful then 14:50:07 eedri: so we at least have some time overview and time to prepare 14:50:11 ewoud, +! 14:50:16 ewoud, +1 14:50:46 #action add jenkins job to email infra@ovirt.org list of unstable jobs before infra meeting 14:52:48 ok 14:52:59 we still got a few min to review tracs 14:53:07 #topic trac tickets 14:53:15 * eedri opening link 14:53:39 https://fedorahosted.org/ovirt/query?status=new&status=reopened&order=priority&col=id&col=summary&col=status&col=type&col=priority&col=milestone&col=component looks promising 14:53:57 #17 implies backups are not being taken? 14:54:08 orc_orc, they are, 14:54:19 orc_orc, jenkins, gerrit are backed up 14:54:27 ewoud, is foreman too ? 14:54:43 ewoud, in $company we have a script that backups foreman db iir 14:54:44 eedri and so, are they tested automatically as to usability? 14:55:01 orc_orc, no, don't think we reached that point yet 14:55:07 * nod * 14:56:43 it would seem puppet or such could be asked to do this -- also fabiand mentioned igor on one of the mailing lists across the weekend 14:57:12 * fabiand lifts his head .. 14:57:19 fabiand: heh 14:57:56 fabiand: use case is testing https://fedorahosted.org/ovirt/ticket/17 backups 14:58:01 what igor can do is spawn VMs to run functional tests and not clutter some other host. it was started ot test rhev but can probably spawn VMs with other OSs too 14:58:34 ah no, igor is not for taking backups :) 14:58:49 orc_orc, testing backips doesn't requires vms or spawning vms, foreman can handle it 14:59:06 ... not for taking -- for doing deployments form such images to see if they are usable 14:59:06 orc_orc: Error: ".." is not a valid command. 14:59:14 \... not for taking -- for doing deployments form such images to see if they are usable 14:59:27 #action check if all critical services are backed-up (jenkins/gerrit is currently backedup) 14:59:29 from* 15:00:17 ok, we don't have much time, but i will say a few words on #73 15:00:38 we're lacking setup/upgrade jobs automation on ci since otopi migration 15:01:13 this is something we need to push forward, and i'm in the process of checking which is the best way to achive it 15:01:33 we might have existing code we can use for it, i will have an answer for next meeting 15:02:02 one last thing - obasan tested nested virt, and it seems to work w/o issues 15:02:28 we should covert one of the existing vms on rackspace to support nested virtualiztion so we can runs tests on them 15:02:38 i.e create vms 15:02:56 #action obasan to add nested virt support to one of rackspace jenkins slaves 15:03:14 eedri: nested virtualization does not work on RHEL 6'ish libvirt as I recall -- so it needs a Fedora host? 15:03:32 orc_orc, yes. the bare metal running the vms is fedora 19 15:03:45 orc_orc, only supported on fedora kernel for now 15:04:04 orc_orc, but on top of it you can crearte centos/rhel/fedora vms 15:04:07 it is a kernel capability needed, not a libvirt one? 15:04:21 orc_orc, afaik its supported in the kernel 15:04:22 obasan, ? 15:04:32 eedri: thanks 15:04:53 ewoud, anything to add before we close? 15:05:00 eedri, nested virt only works on fedora 15:05:28 going once 15:05:46 2 15:06:11 ok, cya everyone, if you have issue. please email infra@ovirt.org 15:06:15 #endmeeting 15:06:46 ewoud, ? 15:07:29 strange that a chair cannot end a meeting 15:07:31 #endmeeting