15:50:00 #startmeeting oVirt Infra 15:50:00 Meeting started Mon Feb 3 15:50:00 2014 UTC. The chair is knesenko. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:50:00 Useful Commands: #action #agreed #help #info #idea #link #topic. 15:50:08 #topic Hosting 15:50:30 #chair ewoud eedri 15:50:30 Current chairs: eedri ewoud knesenko 15:50:33 Rydekull: here ? 15:51:27 #chair obasan 15:51:27 Current chairs: eedri ewoud knesenko obasan 15:51:31 * orc_orc is here 15:51:35 #chair orc_orc 15:51:35 Current chairs: eedri ewoud knesenko obasan orc_orc 15:52:06 anyone from Kimchi project here ? 15:53:55 ok 15:54:00 so what do we have on hosting ? 15:54:08 eedri: want to update us a little bit ? 15:54:27 knesenko, yes 15:54:43 knesenko, well, as you all know we've been looking into migrating to another vendor 15:55:07 knesenko, while current vendor didn't provide the support and performance we expected 15:55:51 knesenko, we currently have an option to move to softlayer, but it seems that with the availiable budgent i'm not sure we'll be able to fully migrate like we want to 15:56:02 hey guys, the admin interface in my ovirt installation doesn't show any VM, the progress bar just doesn't go away, anyone has the same problem with the latest git? or possibly a solution? 15:56:14 knesenko, i.e building a strong and reliable infra to hold all our servers and vms 15:56:47 knesenko, so we're still looking into another possible hosting option that might solve it, but i don't have updates on it yet 15:57:04 eedri: ok 15:57:05 thanks 15:57:13 knesenko, so i'm tending to wait a bit and try to get an answer soon rather than start migration to a new vendor 15:57:17 eedri, are there any special requirements a hosting options needs to fullfill for us? 15:57:21 knesenko, and getting stuck in the middile 15:57:27 #info eedri is looking for a best solution for the infra. no updates yet 15:57:36 fabiand, well, we talked about it quite a bit, but in general what we need is: 15:57:48 fabiand, 1 strong VM to host the ovirt-engine 3.3 stable 15:58:10 fabiand, 3 strong hypervisors to run jenkins slaves with various os's 15:58:36 fabiand, and most important a storage server to support nfs/iscsi to serve both for automation tests and also as storage domain 15:59:08 fabiand, hopefully we can add a production DC over the new engine and migrate all the vm currently on alterway to there as well and mange the alterway servers from it also 15:59:34 fabiand, so we'll end up with a single manager that manages a few DCs 15:59:39 eedri, okay - thanks for the summary 15:59:42 knesenko, yeap! I am from kimchi 15:59:47 fabiand, i'll also migrate the jenkins master into a vm once we can 15:59:49 knesenko, sorry for the late message 15:59:54 alinefm: hey 16:00:04 alinefm: want to join our meeting / 16:00:05 ? 16:00:17 knesenko, sure 16:00:18 alinefm: and after we can discuss the details what do you need etc. ? 16:00:22 #chair alinefm 16:00:22 Current chairs: alinefm eedri ewoud knesenko obasan orc_orc 16:00:40 alinefm: would you like to introduce yourself ? 16:01:16 knesenko, well, my name is Aline. I am Kimchi maintainer 16:01:29 Kimchi is a web-based application to manage virtual machines 16:01:39 https://github.com/kimchi-project/kimchi 16:02:06 alinefm: great ... ! 16:02:09 thanks ... 16:02:15 lets continue guys 16:02:20 so anything else on hosting ? 16:02:38 ok 16:02:50 #topic Foreman and Puppet 16:03:03 ewoud: hello ! how are you ? any updates for us ? 16:03:22 obasan: ? 16:03:42 knesenko, no updates from my end due to lack of time :( 16:04:07 ewoud: here ? 16:04:53 #info No updates for foreman and puppet 16:05:13 #action we need to work on slaves puppet classes to configure repos etc. 16:05:19 #topic Jenkins 16:05:32 eedri: obasan want to update us a little bit ? 16:06:33 knesenko, well 16:06:44 knesenko, i think dcaro mostly can update but since he's not around, i'll try to fill in 16:06:57 knesenko, there has been a very disturbing jenkins issues in the past weeks 16:07:02 eedri: I will add my notes as well 16:07:13 knesenko, mostly due to shifting to 3.4 and a high load of patches sent to master 16:07:38 knesenko, in addition, network problems on tlv site which hosts all the minidells caused slaves to disconnect and fail tests 16:07:44 knesenko, blocking patches from being merged 16:08:07 knesenko, solution was to 1st disable some of the heavy jobs like findbugs per patch 16:08:30 knesenko, second the executors on the minidells were reduced to 2, to avoid context switch an overloading the servers 16:08:47 knesenko, also, dcaro installed ssh tunneling which improved the connectivity of the hosts 16:09:13 knesenko, in general status is much stable now and better, though i think there are more slaves on rackspace that needs reprovision 16:09:18 eedri: so now all slaves from tlv site are using ssh tunnels ? 16:09:24 knesenko, afaik, yes 16:09:26 instead of java ? 16:09:29 eedri: ok ... 16:09:33 knesenko, need to sync with dcaro 16:09:46 knesenko, in addition lots of jobs were cloned to run on 3.4 stable branch 16:09:51 #info dcaro updated the tlv slaves to use ssh tunnels instead of java 16:09:53 knesenko, so we have now 3 versions supported 16:09:57 knesenko, 3.3/3.4/master 16:10:19 #info jenkins support 3 versions 3.3/3.4/master=3.5 16:10:22 great 16:10:27 eedri: finished ? 16:10:28 knesenko, also space was increased on resources.ovirt.org to support new rpms 16:10:41 knesenko, i think we need to upgrade jenkins asap 16:10:45 knesenko, to latetest LTS 16:10:50 knesenko, and also update some plugins there 16:10:51 #info space was increased by quaid on resources.ovirt.org 16:11:03 obasan: want to do that ? 16:11:13 knesenko, sure 16:11:30 #action obasan upgrade jenkins server to latest LTS. send email before the outage 16:11:48 obasan, please sync with dcaro and notify list before doing so 16:12:03 eedri, +1 16:12:07 few updates from my side now 16:12:21 I am working hard to support a new layout for the repos 16:12:52 #info new git repo was created for release engineering purposes releng-tools 16:13:14 #info mom project was moved to autotools 16:13:44 #info new publishers were added to support 3.3 and master nightly 16:14:13 #info each job under packaging tab will collect a metadata for the job 16:14:38 #action knesenko prepare 3.4 jobs under packaging tab 16:15:05 I hope a new layout will be announced soon 16:15:13 that's all from my side 16:15:15 questions ? 16:15:45 knesenko: there are a bunch of tools for repository management in yul-utils as I recall -- will the new layou6t integrate with those? 16:15:51 yum-utils* 16:16:18 orc_orc, i think knesenko code manages the layout of the dirs and copying of files 16:16:19 they handle lifecycle, closure, and such 16:16:30 orc_orc, less on repository mgmt, which afaiu is a post phase of it 16:16:37 orc_orc, and maybe we can add it once it's ready, knesenko ? 16:16:51 orc_orc, also maybe implementing a UI interface to it 16:16:59 orc_orc: I didn't looked into that ....mostly I worked on pusblishing rpms from nightly jobs to resources.voirt.org 16:17:16 orc_orc: but I am pretty sure I need to take a look on yum-utils ... 16:17:44 but that's will be my next step 16:18:07 knesenko: ask if you need help as I am familiar with them and it does simplify automated lifecycle 16:18:13 knesenko, orc_orc, i think once we have it ready, we can look into managing the repos with an existing utility, which was suggested by sbonazzo or brian in the list 16:18:34 #action talk to orc_orc regarding yum-utils tools to manage repos on a new layout 16:18:41 orc_orc: be sure I will ... soon :) 16:20:01 anything else ? 16:21:19 ok 16:21:39 I think we are done here .... a lot of updates and a lot of work to be done 16:22:09 knesenko, what about kimchi 16:22:26 eedri: I am on it ... let me understand first what is needed 16:24:02 alinefm: :) 16:24:25 eedri, knesenko, we would like to have access to a jenkins server to create jobs for kimchi 16:24:30 nightly builds, builds before applying patches from mail list, ect 16:24:36 I know ovirt already has a jenkins server, maybe we could have a project for kimchi there 16:24:57 alinefm: we are limited in resources .... 16:25:11 do you have machines that we can add as slaves ? 16:25:56 knesenko, I can try to get ones 16:26:11 alinefm: that will be great ... 16:26:19 they need to have a specific configuration? 16:26:34 I think I saw something related to that on ovirt wiki 16:26:40 alinefm: no ... regular os el6 or fedora19/20 16:26:48 alinefm: we will handle the configuration ... 16:27:07 alinefm, what we usually do is give project maintainers "power user" access to add jobs for the project on jenkins 16:27:07 alinefm: or one of your guys :) up to you 16:27:17 alinefm, but like knesenko said, we're kind of low on reources right now 16:27:28 alinefm, maybe as first step you can introduce yourself on the list 16:27:47 alinefm, and talk about the kimchi project and what do you expect from jenkins/reousrces needed etc.. 16:28:21 eedri, sure - I will do that as well 16:28:32 alinefm, so once we'll understand what are facing we can plan it better and allocate resources 16:28:45 knesenko, eedri, do you have slave per projects? 16:28:53 #action alinefm send introduction email to infra@ovirt.org 16:29:03 alinefm, in addition, you're more than welcome to join the infra team on a regular basis 16:29:09 alinefm: we have slaves that serves all jobs 16:29:13 all projects 16:29:20 alinefm, we don't host too many projects that are seperate from ovirt 16:29:21 we are using labels for slaves ... 16:29:34 alinefm, most of the jobs are testing core ovirt components like ovirt-engine & vdsm 16:29:49 alinefm, all other jobs are mainly for packaging, except ovirt-node jobs 16:29:54 each OS has its own label ... 16:29:55 alinefm, which has been there from the start 16:30:21 right, because more than fedora and rhel, kimchi also supports ubuntu and opensuse 16:30:38 in that case, I think a virtual machine with those OS can work 16:30:47 but more you already had this scenario 16:31:00 *but not sure 16:31:01 alinefm: eedri I am ending the meeting 16:31:05 #endmeeting