15:06:41 #startmeeting oVirt Infra 15:06:41 Meeting started Mon Jan 6 15:06:41 2014 UTC. The chair is knesenko. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:06:41 Useful Commands: #action #agreed #help #info #idea #link #topic. 15:06:55 #chair obasan 15:06:55 Current chairs: knesenko obasan 15:06:58 #chair eedri 15:06:58 Current chairs: eedri knesenko obasan 15:06:58 * orc_orc is here 15:07:17 #chair orc_orc 15:07:17 Current chairs: eedri knesenko obasan orc_orc 15:07:19 ewoud: here ? 15:07:24 Rydekull: here ? 15:07:31 bkp: here ? 15:08:17 knesenko: Yes 15:08:24 #chair bkp 15:08:24 Current chairs: bkp eedri knesenko obasan orc_orc 15:09:08 orc_orc, i have a conf live if you want to join a voice chat to discuss the current stats 15:09:17 kkwill call in 15:09:22 bkp: ^^ 15:11:26 #topic Hosting 15:11:57 Hi all 15:12:11 I hope you enjoyed your holidays 15:12:13 not call ID: 842-597-391-5 ? 15:12:45 knesenko: Monitoring, but on another call. 15:12:52 bkp: ok 15:12:54 orc_orc, the bridgeid: 1959353860 15:13:10 * nod * 15:13:12 so I installed monitoring.ovirt.org/cacti service to monitor our infra 15:13:21 and we have few stats for it ... 15:13:39 you can find the credentials in a usual place 15:13:51 #info knesenko monitoring.ovirt.org/cacti was installed 15:16:19 #info http://etherpad.ovirt.org/p/Hardware_requirements_infra 15:20:07 knesenko: semi here, highlight if needed 15:20:22 ewoud: we are on the phone call 15:20:24 ewoud: ok 15:20:54 knesenko: ok, will read the minutes 15:23:05 http://www.softlayer.com/dedicated-servers/dual-processor-servers 15:26:13 Hi infrateam, just a info from your user: the main ovirt.org repo is very slow regarding downloading packages, I never get more than 250kbps which leads to some time to get a new setup running, any chance of speeding this up? 15:30:49 Is the etherpad chat disabled? 15:33:07 SvenKieske: I do not see it either 15:34:05 well, nevermind, I just did point to my message from here, regarding the downloadspeed from the main repo. does nobody else got a problem with that? I already reported that some months ago via ML but never got any replies :( 15:34:48 statistics: sysstat 15:35:12 adding a wrapper to a Jenkins process start to log load 15:35:27 also, bail on a new jenkins when load too high 15:35:41 is there a re-queue mechanism> 15:37:21 #action try to use sysstat for our server to get more statistics 15:37:58 #action look for a load management plugin for jenkins and slaves 15:38:48 w | logger -p local1.info 15:39:30 w | top -n 1 and check the value 15:39:38 #action try to change vCpu allocation per slave to see if that affects the cpu load on rackspace servers 15:39:51 w | head -n 1 and check the value 15:39:59 10:39:56 up 48 days, 5:29, 39 users, load average: 1.64, 1.64, 1.42 15:40:17 if one lf last 3 too bi, abail and ask for a wrequeue 15:40:20 hi* 15:44:20 #info we found from the current statistics that our hypervisors mostly have high CPU 15:45:31 #info we agreed to reduce the vCPU on slaves to see how it will reduce the load 15:46:03 #info we will try to gather more statistics with the tools orc_orc provided 15:46:16 ok I think we summarized most of the issues here in hosting 15:46:25 I am moving to another topic 15:46:38 #topic 15:46:48 #topic Foreman and Puppet 15:46:54 ok what do we have here 15:47:29 we have few patches here waiting for merge 15:47:37 http://gerrit.ovirt.org/22761 15:47:42 http://gerrit.ovirt.org/22762 15:48:29 ewoud: can you review them ^^ ? 15:48:42 we need cacti user for reverse ssh tunnel ... 15:49:00 because we have few slaves behind a NAt 15:49:54 moving to jenkins 15:50:00 #topic Jenkins 15:50:05 eedri: helo :) 15:50:17 eedri: ok ok ok 15:50:26 so I did some work here as well 15:50:51 #info added missing builders for some projects http://jenkins.ovirt.org/view/Packaging/ 15:51:11 I added new publishers to resources.ovirt.org 15:51:21 now we will have publisher per version 15:51:30 #info new publishers http://jenkins.ovirt.org/view/Publishers/ 15:51:48 Also I am working on a new layout for repos 15:52:02 #info http://resources.ovirt.org/pub/ 15:52:08 example for the new layout 15:52:15 each version will land in its place 15:52:32 knesenko: there was concern on the ML that hits will bread 'downgrades' under yum? 15:52:47 orc_orc: it will work with a new layout 15:53:07 can a 'flat' all in one directory view be made, using the 'hardlink; program, as well 15:53:12 ok ... 15:53:18 how to test that it is working? 15:53:44 because the 'repodata' directories will be down a lot of levels 15:53:53 what do you mean ? 15:54:27 the new layout has a vertical string of directory and sub directories one moves through 15:54:37 ok 15:54:39 to reach the repodata related toa given release point 15:54:54 but sometimes a person wants to 'fall back' to an earlier working version 15:55:04 orc_orc: got you 15:55:06 which yum supports through the ... 15:55:18 downgrade command 15:55:34 and for that to work, yum needs to see the eralier versions in the repodata 15:55:35 lets take for example - http://resources.ovirt.org/pub/ovirt-3.3-snapshot/rpm/Fedora/20/ 15:55:41 you are running fedora 15:55:54 all nightly 3.3 builds will land there 15:56:21 and we will save for example 'n' builds 15:56:32 so you will be able to downgrade 15:57:01 ... so will the prior versions be down in a directory under http://resources.ovirt.org/pub/ovirt-snapshot/rpm/Fedora/20/ with its own repodata? 15:57:01 orc_orc: Error: ".." is not a valid command. 15:57:04 ... so will the prior versions be down in a directory under http://resources.ovirt.org/pub/ovirt-snapshot/rpm/Fedora/20/ with its own repodata? 15:57:31 (actually both the latest and the prior versions, I guess) 15:57:42 ovirt-snapshot - is for master builds 15:57:57 master is the next version 15:58:18 so where will the union of 3.2 and 3.3 content be under your model? 15:59:00 http://resources.ovirt.org/pub/ovirt-${version}-snapshot/ 15:59:12 -snapshot - nightly builds 15:59:30 http://resources.ovirt.org/pub/ovirt-${version} - released builds 15:59:47 http://resources.ovirt.org/pub/ovirt-snapshot - master nightly builds 16:02:00 orc_orc: I will send an email on it 16:02:04 ok guys thanks 16:02:08 knesenko: thank you 16:02:11 I think we are done here 16:02:18 orc_orc: thanks for help ! was nice talking to you 16:02:23 indeed 16:02:26 #endmeeting