14:24:51 #startmeeting infra weekly 14:24:51 Meeting started Mon Jun 3 14:24:51 2013 UTC. The chair is ewoud. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:24:51 Useful Commands: #action #agreed #help #info #idea #link #topic. 14:25:01 #chair eedri knesenko dcaro obasan 14:25:01 Current chairs: dcaro eedri ewoud knesenko obasan 14:25:23 a bit later than planned because I had to help a colleague 14:25:43 ewoud: ok 14:26:55 #topic hosting 14:27:35 #info rackspace01.ovirt.org installed 14:27:38 I'm around! 14:27:43 dcaro, hey 14:27:44 dcaro: rackspace02 as well? 14:27:48 yep 14:27:59 they are not connected to foreman yet, not have puppet 14:28:13 but the base os is installed and all the admins have access 14:28:25 (to keep setting it up) 14:28:26 #undo 14:28:26 Removing item from minutes: 14:28:45 #info rackspace0{1,2}.ovirt.org installed, not connected to foreman/puppet yet 14:29:08 dcaro: any specific problem with puppet or just lack of time? 14:29:56 ewoud: some problems with the ruby version 14:30:13 (and lack of time to solve it) 14:30:34 dcaro: ok 14:31:08 dcaro: is ovirt installed on them? 14:31:40 not yet asfaik, knesenko: ^ 14:31:46 ewoud, not yet, we want to rethink the infra deployment before installing it 14:31:55 eedri: ok 14:32:02 ewoud, might not be optimal to do it as we planned, knesenko can elaborate 14:32:45 knesenko: I heard you could elaborate why we want to rethink the infra deployment 14:32:48 knesenko, can you share some insight on your plan for the infra of ovort-engine 14:33:06 knesenko has previous experiene with ovirt-engine installations 14:33:11 yes ... It make no sense to install alinone setups 14:33:40 so we need to define one engine that will manage our hosts ... for now we have 4 bare metal . 14:34:05 so what we need is - one amazon VM - for engine and it will manage our hosts. 14:34:20 this is the correct architecture ... 14:34:37 but for this we will need storage ... some external storage ... NFS or iscsi 14:34:46 knesenko, we'll need to check what is the overhead (if exists) to manage multiple hosts access WAN/internet 14:35:28 the only problem that we need a storage domains that will be located near the hosts ... 14:36:07 dneary, is it possible to get storage service from rackspace? 14:36:32 eedri, I hate to say "I don't know", I haven't dealt with RAX at all 14:36:41 dneary, will make our mgmt of infra much simpler and resilient than using one host as storage server 14:37:08 dneary, ok. we'll have to query karsten then 14:37:22 eedri, Would Akamai be an option? 14:37:46 dneary, i think any storage service that can provide iscsi/nfs and open to the hosts will do 14:37:50 brb 14:38:17 knesenko, can you send a desc of the proposed infra layout to the list? including request for the storage service 14:38:50 eedri: of course ... need to prepare it first 14:38:54 knesenko, np 14:39:16 #info knesenko to send desc of ovirt-engine layout to infra list 14:39:26 hi, im having dependency issues with some packages in centos 6, 14:39:37 vdsm-gluster-4.10.3-10.el6.noarch seems to be outdated 14:39:51 BlairQ, we're in the middle of ovirt meeting, can you wait till it's over please? 14:40:04 off course, im sorry 14:40:11 BlairQ, np 14:40:15 BlairQ: its ok :) 14:40:15 ewoud, continue? 14:40:41 he is not here ... 14:40:44 lets go ... 14:40:51 ok, what else do we have on hosting 14:41:19 obasan: incinga ? 14:41:34 knesenko, I have a puppet class to manage nrpe 14:41:46 knesenko, but as dcaro mentioned it does require tweaking 14:41:53 knesenko, has to be authenticated 14:42:08 https://github.com/Ohadbasan/puppet-nrpe 14:42:32 obasan: authenticated ? 14:42:43 * ewoud is back 14:42:49 knesenko, ssl between the icinga server and the monitored hosts 14:42:57 nrpe by default is not secured 14:42:59 ewoud, i think we're done with hosting, maybe we should change topic to puppet? 14:43:19 eedri: feel free to do it yourself next time when I'm not around :) 14:43:22 #topic puppet 14:43:52 ewoud, i was about too :) 14:44:21 obasan: I did add the hosts manually to icinga to get a good feeling with it 14:44:35 but I'd really like to automate it 14:44:57 ewoud, https://github.com/Ohadbasan/puppet-nrpe 14:45:29 ewoud, auto adding of hosts can also be done 14:45:37 obasan: I was looking at it, will do a pull request with what I think is some beautification 14:45:51 also, I don't like ensure => latest in puppet 14:46:08 ewoud, feel free to commit changes if you wish :) 14:46:21 obasan: more parameters would be better :) 14:46:28 is anyone else having a problem doing an update today using ovirt-stable? 14:46:49 i am getting i686 packages where I had none before (and they are causing a failure) 14:46:56 ewoud, nrpe.cfg is pretty static... 14:47:28 Error: Package: vdsm-4.10.3-10.fc18.i686 (ovirt-stable) 14:47:34 yes .... again... vdsm 14:48:04 but it seems the x86_64 version is trying to (for some reason) force the install of i686 multilib stuff 14:48:23 i don't have any i686 stuff installed 14:49:04 and --skip-broken does not fix it 14:49:20 nor does --setopt=protected_multilib=false 14:49:30 Protected multilib versions: vdsm-python-4.10.3-16.fc18.x86_64 != vdsm-python-4.10.3-10.fc18.x86_64 14:50:25 again... ihave nothing i686 installed and it is trying to install i686 stuff 14:51:00 xoritor, can you please post these questions in 30 min? we're currently in an ovirt meeing 14:51:10 sure 14:51:40 ewoud, we're still waiting for approval for puppet repo on gerrit.ovirt.org right? 14:51:42 BlairQ: xoritor I will try to help you both after the meeting 14:52:12 * quaid doesn't know off-hand of storage from RackSpace 14:52:12 thanks... sorry didn't read the /topic ... sorry 14:52:20 eedri: yes 14:52:27 obasan: at $employer we use https://github.com/ekohl/nrpe_basic 14:52:41 which is rather basic as well, but does a bit more 14:53:25 quaid, can we open a ticket in rackspace to get info on that? 14:53:36 ewoud, seems to be ok 14:54:02 eedri: sure, any of us can ask, but it's probably ultimately something to talk with Jim Strong about 14:54:06 ewoud, it just supports all sorts of linux platforms. 14:54:56 quaid, ok, let's wait till knesenko will send his proposal for the ovirt-engine infra layout and then send our query to jim (with all the options/alternatives) 14:55:27 sorry, I really wish I had more time for the meeting today 14:55:32 #chair quaid 14:55:32 Current chairs: dcaro eedri ewoud knesenko obasan quaid 14:55:45 ewoud: np 14:55:52 #info puppet repo on gerrit requested 14:56:17 how do we want to push to foreman? using gerrit replication or a jenkins job? 14:56:48 dcaro, i think you have experience with that, right? 14:57:14 does gerrit have proper logging of the replication? 14:58:06 right now we are doing it with a gerrit plugin @redhat, but a jenkins job should work also ok 14:58:07 I know jenkins can log easily 14:58:41 but I'd be fine with gerrit as well if we can get debug output when it fails 14:58:43 or a notification 14:58:47 *gerrit hok sorry 15:00:04 maybe we can link it to an icinga passive check, and update it from the hook 15:00:47 ok, let's move on the topics and continue this on the list? 15:00:47 why is a jenkins job with gerrit trigger a bad idea? 15:00:51 it seems like the easiest solution 15:01:00 we're out of time for the meeting and still got a lot to cover 15:01:05 once you commit to puppet repo > jenkins job runs :) 15:01:10 eedri, ok 15:01:49 #action continue puppet repo push to foreman on infra list 15:02:06 #topic jenkins 15:02:13 I've had a bit of a monologue on the puppet repository layout, I'd appreciate if you could reply to it btw 15:02:20 ewoud, +1 15:02:42 ewoud, sorry i didn't have time to go over it, been very busy in $employer lately 15:02:55 shame on me, I have it pending since you sent it :S, sorry+ 15:03:19 i added a new job in jenkins for running find_bugs per patch on ovirt-engine 15:03:41 the tab is still open on my browser :) 15:03:48 i know this is not optimal in current infra and we wanted to wait till new vms are alive 15:03:50 will go over it asap 15:03:54 obasan: so reply to the email ! 15:04:09 but lately there has been too many failures on the jenkins on compilation and findbugs.. 15:04:30 it seems jenkins is coping with the load for now, but it might be it's overloading the gerrit instance and causing it to fail occasionaly 15:04:53 itamar is planning to upgrade the gerrit instance to 2.5, hopefully it will solve the issue 15:05:22 i will track the status of jenkins for the next few days, if this erratic behavior continues, i'll disable the job for now 15:05:51 * eedri finished 15:06:35 I propose adding someone more to be able to manage gerrit (with admin rights to the instance), so Itamar does not have to be on it fridays too 15:06:38 another jenkins jobs in progress is testing engine-setup + engine-upgrade, which is failing on real issues 15:07:01 dcaro: +1 15:07:01 dcaro, +1 15:07:05 dcaro, +1 15:07:33 dcaro: +1 15:08:17 #action add an additonal administrator for gerrit.ovirt.org to add redundancy and backup for weekends 15:08:47 any other jenkins related stuff? 15:09:10 quaid, i hope you're around cause other than changing topic i don't know how to manage the bot :) 15:09:53 eedri: I'm here 15:09:59 finished backup for jenkins 15:10:04 right 15:10:07 thanks knesenko 15:10:27 #agreed Add an additional admin for gerrit 15:10:29 :) 15:10:35 * quaid tries out some other things 15:10:46 #info jenkins configuration files are now backed up to alterway02 server 15:11:05 nightly backus, kept for 30 days 15:11:25 we should consider a full backup of jenkins homedir, around once a month... 15:11:35 prefferably to a remote location (rackspace?) 15:11:56 maybe we will have some ext. storage 15:12:03 lets wait with a full backup for now 15:12:06 not just jenkins, other ciritical servers as well 15:13:00 #action re-discuss full backup of infra services once we have ext. storage 15:13:21 we can now yum upgrade alterway01 (jenkins.ovirt.org) 15:14:13 #info added new job to run quick rpm make for ovirt-engine on fedora and el (to verify packging isn't broken) 15:14:18 ok, moving on 15:14:51 do we have time to review trac tickets 15:14:52 ? 15:15:16 eedri: hmmm ... have no idea . 15:15:24 ewoud, quaid dneary ? 15:16:04 i think we're overdue the meeting time anyhow, we'll do it next time 15:16:05 seems like 50%+- are assigned to a specific users 15:16:52 ewoud, ping 15:17:01 ewoud, do you want to add action to reschdule meeting time? 15:18:00 Guys , I would like to add an oVirt company proifile to linkedin same as Fedora - http://www.linkedin.com/company/32178?trk=vsrp_companies_res_name&trkInfo=VSRPsearchId%3A528744561370272644773%2CVSRPtargetId%3A32178%2CVSRPcmpt%3Aprimary 15:18:32 but for this purpose I need .*@ovirt.org email account . is it possible to open an account for linkedin or something ? 15:18:35 quaid: ^^ 15:18:39 knesenko, i suggest bringing it up the list 15:18:55 eedri: I will 15:19:07 knesenko, good idea 15:19:32 #action reschdule infra meeting time after reviewing doodle results 15:19:45 please respond to ewoud email on changing meeting time 15:20:05 eedri: ok 15:20:11 end meeting ? 15:20:19 i think we're ready to close the meeting unless someone has something to say? 15:20:35 going once 15:20:41 twice... 15:20:50 sold! 15:20:51 quaid, .. can you end the meeting? 15:20:54 I object! 15:20:55 :) 15:21:32 #action send request to add ovirt as project to linkedin 15:21:32 knesenko: we can create an alias or a marketing list or something, too 15:21:45 * quaid trie the meeting end trick 15:21:46 quaid, +1 15:21:53 #endmeeting