16:00:21 <xarses> #startmeeting fuel 16:00:23 <xarses> #chair xarses 16:00:23 <openstack> Meeting started Thu Aug 6 16:00:21 2015 UTC and is due to finish in 60 minutes. The chair is xarses. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:24 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:25 <xarses> Todays Agenda: 16:00:26 <openstack> The meeting name has been set to 'fuel' 16:00:27 <xarses> #link https://etherpad.openstack.org/p/fuel-weekly-meeting-agenda 16:00:28 <openstack> Current chairs: xarses 16:00:29 <xarses> Who's here? 16:00:31 <mattymo_> hi! 16:01:14 <xenolog13> \~/ 16:01:19 <angdraug> o/ 16:01:20 <mihgen> hi 16:01:25 <akasatkin> hi 16:01:43 <vkramskikh> hi 16:01:46 <rvyalov> hi 16:01:48 <amaksimov> hi 16:02:02 <xarses> ok, lets get started 16:02:05 <xarses> #topic SCF Status (angdraug) 16:02:11 <angdraug> 7.0 soft code freeze will be in effect at 12AM tonight 16:02:11 <angdraug> today is the last day to merge anything that is not a fix for High or Critical bug 16:02:11 <angdraug> we're still in the red zone with the number of bugs 16:02:20 <angdraug> dixi :) 16:02:50 <angdraug> questions? moving on? 16:02:54 <mihgen> angdraug: rationale is to focus on high and critical now, right? 16:03:00 <angdraug> yes 16:03:03 <amaksimov> question: 16:03:04 <mihgen> so to converge by 3rd of Sep 16:03:26 <amaksimov> we will not merge anything related to Med bugs after SCF,right ? 16:03:32 <angdraug> yes 16:03:35 <mihgen> angdraug: and I think if there is minor work to be completed in some low / medium bugs in progress, those can still be merged after SCF 16:03:43 <mihgen> let's say for another week 16:03:52 <amaksimov> ok 16:04:07 <amaksimov> our review stat might affected 16:04:13 <angdraug> if there is medium bug or lower priority related work still pending, now is the time to speak up and ask for merge or exception 16:04:29 <xarses> mihgen: how will we accomidate that? SCF exceptions? 16:04:50 <angdraug> reviews affected by code freeze should be marked workflow-1 until HCF when we create stable/7.0 branches 16:05:05 <vkramskikh> what if there is a review request which closes 1 high and 2 medium bugs? 16:05:22 <angdraug> vkramskikh: that doesn't need an exception 16:05:34 <angdraug> as mihgen said, the motivation is to focus on fixing high bugs 16:05:35 <amaksimov> this is not going to work.. we will launch auto abandon script which will abandon reviews with -1 16:05:59 <xarses> amaksimov: its not supposed to hit workflow -1 16:06:27 <xarses> If some one sees it do so, please speak up so we can fix the script 16:06:33 <amaksimov> sure? I think we need to check it with devops. 16:07:05 <amaksimov> tiran ? 16:07:09 <mihgen> So I would say there are two major things with SCF: 1) to focus on High/ Criticals to meet deadlines; 2) to lower the risk of introducing regressions by fixes of low/medium. The less code change is, the lesser risk is 16:07:13 <angdraug> #link https://review.fuel-infra.org/#/c/10080/3/requests-abandon/abandon_old_reviews.sh 16:07:13 <mihgen> teran: ^^ 16:07:33 <mihgen> amaksimov: script should not touch workflow -1, if it does - then it has to be fixed 16:07:39 <angdraug> it only abandons reviews with -2 16:08:06 <teran> mihgen: angdraug AFAIR it works already such way 16:08:11 <teran> we could check it out 16:08:21 <angdraug> yes, that's my impression as well 16:08:38 <angdraug> amaksimov: does that answer your question? 16:08:42 <amaksimov> yes 16:08:43 <akasatkin> some mediums just need to be checked. what to do with them? 16:08:55 <mihgen> akasatkin: like what? 16:09:12 <akasatkin> https://bugs.launchpad.net/fuel/+bug/1399004 16:09:12 <openstack> Launchpad bug 1399004 in Fuel for OpenStack "Running network verification on running env may break network" [Medium,Confirmed] - Assigned to Fuel Python Team (fuel-python) 16:09:14 <uvirtbot> Launchpad bug 1399004 in fuel "Running network verification on running env may break network" [Medium,Confirmed] 16:09:15 <uvirtbot> Launchpad bug 1399004 in fuel "Running network verification on running env may break network" [Medium,Confirmed] https://launchpad.net/bugs/1399004 16:09:33 <mihgen> this sounds rather High 16:09:45 <akasatkin> okay 16:09:45 <angdraug> +1 16:10:15 <xarses> ya, thats high 16:10:19 <mihgen> break of deployment in general sounds critical issue, if it can be reproduced only in certain cases - then it's high 16:11:05 <amaksimov> it doesn't break deployment 16:11:09 <amaksimov> just verification 16:11:12 <amaksimov> sometimes 16:11:14 <angdraug> to reiterate, the reason we're announcing soft code freeze earlier is to spend more time on high and critical bugs 16:11:49 <mihgen> amaksimov: nope according to the bug, you break working env 16:11:53 <mihgen> you load kernel module 16:11:56 <mihgen> and ovs breaks 16:12:10 <mihgen> moving on? 16:12:11 <xarses> ok, lets move along 16:12:15 <xarses> #topic puppet-librarian (angdraug) 16:12:50 <angdraug> puppet-librarian commits are being merged according to plan: 16:12:50 <angdraug> #link http://lists.openstack.org/pipermail/openstack-dev/2015-August/071347.html 16:12:50 <angdraug> to honor soft code freeze, I propose to merge the firewall module one day early 16:12:50 <angdraug> cinder module will need a rebase over https://review.openstack.org/209945 16:12:50 <angdraug> we should discuss an SCF exception for it next week 16:13:07 <angdraug> thoughts? objections? 16:13:29 <xarses> I think its minor, and we need to move forward with it 16:13:30 <xarses> +1 16:13:52 <angdraug> #link https://wiki.openstack.org/wiki/Fuel/Library_and_Upstream_Modules 16:13:55 <mattymo_> We had 1 custom patch to firewall module for docker stuff, but that isn't relevant today 16:14:12 <mattymo_> actually... now that I think about it, it was just a backport of a more modern version 16:14:24 <mattymo_> carry on, please continue with firewall module 16:14:29 <angdraug> thanks! 16:14:57 <angdraug> apache is already merged, this means we'll also merge apt and firewall today, and leave cinder until next week 16:15:17 <angdraug> that's it from me 16:15:20 <xarses> thanks 16:15:24 <mihgen> angdraug: cinder until HCF 16:15:36 <sgolovatiuk> Hi Fuelers... 16:15:57 <angdraug> mihgen: most likely, unless there's strong consensus in favor of SCF exception 16:16:01 <xarses> #topic Open discussion (bugs, reviews that need attention) 16:16:06 <mihgen> If we enter SCF I'm not sure if we want to continue doing exceptions of any kind 16:16:28 <amaksimov> if we want to release on time - no exceptions please 16:16:33 <angdraug> fair enough 16:16:40 <mattymo_> xarses, now that we're trying to focus on high priority bugs, we should make an effort to not "hog" all the bugs 16:16:41 <mihgen> amaksimov: +1 16:17:02 <mattymo_> If you hold a high priority bug and you're not working on it, pass it back to fuel-{python,library,etc} 16:17:08 <mattymo_> one that needs to be fixed for 7.0 16:17:12 <mattymo_> that way others can pick them up 16:17:14 <xarses> mattymo_: are we having problems with bugs assigned and not being worked again? 16:17:21 <amaksimov> yes 16:17:25 <amaksimov> +1 mattymo_ 16:17:37 <mattymo_> our statistics are bad, but the # in fuel-library queue are not so high 16:17:45 <mihgen> mattymo_: +1 16:18:09 <angdraug> can we get a report on number of assigned open bugs per person? 16:18:12 <mihgen> we need to do daily triage passes over all bugs left 16:18:16 <IvanKliuk> https://bugs.launchpad.net/fuel/+bug/1365368 is tricky. I've been trying to fix it but I'm facing the problems with syncronization. 16:18:16 <openstack> Launchpad bug 1365368 in Fuel for OpenStack 7.0.x "Cannot update ip_ranges for management and storage networks" [High,In progress] - Assigned to Ivan Kliuk (ivankliuk) 16:18:16 <uvirtbot> Launchpad bug 1365368 in fuel/7.0.x "Cannot update ip_ranges for management and storage networks" [High,In progress] 16:18:17 <uvirtbot> Launchpad bug 1365368 in fuel/7.0.x "Cannot update ip_ranges for management and storage networks" [High,In progress] https://launchpad.net/bugs/1365368 16:18:46 <mattymo_> mihgen, triage of all bugs owned by individuals? that's a tricky challenge. better to solve it in a bug squashing session where we can get everyone together 16:18:53 <amaksimov> angdraug in our KPI report 16:19:11 <mihgen> mattymo_: I mean get everyone together, and go over all bugs 16:19:16 <mattymo_> +1 16:19:32 <xarses> IvanKliuk: wha't ca we do to help? 16:20:01 <xarses> It seemed to work fine for networks of type ip_range, doesn't work for type 'cidr' 16:20:15 <IvanKliuk> xarses: No help needed, thanks 16:20:20 <angdraug> mihgen: "going over" 100 bugs is going to take all day, doing it with the whole team will take up all day of the whole team, not sure about that 16:20:48 <mattymo_> angdraug, we manage to do it every release, usually twice :) 16:20:53 <IvanKliuk> I just cannot estimate when it's going to be fixed 16:21:02 <xarses> ok 16:21:09 <IvanKliuk> because it's not trivial fix 16:21:29 <xarses> any one else have CR or bugs to discuss? 16:21:35 <mihgen> I managed it before in this way: I was going over bugs, and doing ping-pong of folks working in the area 16:21:42 <IvanKliuk> I'll let you know about the results by the end of the day via email 16:21:51 <mihgen> to whether provide update, ask if they work on a bug or not, etc. 16:22:23 <angdraug> I like that approach better than "get everyone together":) 16:22:35 <mihgen> however one or two passes by whole team are still needed 16:22:51 <mihgen> to see the whole picture and hear concerns from everyone 16:22:55 <angdraug> amaksimov: found it, thanks. we can use that to identify who's hogging bugs and make sure they don't just sit on any of them 16:22:59 <mihgen> around certain bugs 16:23:30 <angdraug> sounds like a good agenda for the next IRC meeting 16:23:41 <mihgen> how are we doing with upgrades? 16:23:55 <angdraug> I hear last bits were merged yesterday 16:24:01 <amaksimov> yes 16:24:09 <amaksimov> I talked with ogelbukh today 16:24:18 <amaksimov> all ISO related stuff were merged 16:24:42 <amaksimov> but scripts still not ready 16:24:54 <amaksimov> but they are not part of ISO 16:25:22 <mihgen> ok.. 16:25:38 <mihgen> at least we merged all ISO-related code 16:25:58 <amaksimov> yes 16:26:00 <mihgen> what about centos 6.6 for master node support? 16:26:08 <amaksimov> not yet merged 16:26:30 <angdraug> staging was blocked earlier today 16:26:38 <angdraug> as soon as it's green we can merge it 16:27:02 <mihgen> why was it red? 16:27:03 <angdraug> the cinder commit I mentioned earlier is expected to fix staging bvt: 16:27:17 <angdraug> #link https://review.openstack.org/209945 16:27:31 <angdraug> a regression introduced by a cinder package 16:28:05 <mihgen> ok 16:28:26 <mihgen> so do we have deadline set for 6.6 ? 16:28:58 <angdraug> today 16:29:15 <amaksimov> btw shouldn't we discuss the necessity to merge centos 6.6 ? 16:29:33 <angdraug> amaksimov: didn't we already have enough discussions about it? 16:29:33 <amaksimov> we had several conversations inside 16:29:46 <amaksimov> maybe we also should discuss it here? 16:30:04 <angdraug> it's not a fuel change, it's a mos change 16:30:22 <mihgen> it's Fuel relevant I'd say 16:30:31 <mihgen> Fuel itself has to support it 16:30:34 <amaksimov> we will update kernel 16:30:57 <mihgen> do we update bootstrap as well? 16:31:13 <angdraug> yes 16:31:17 <amaksimov> upgrade script will become more complicated 16:31:29 <amaksimov> it will require reboot of master node 16:31:35 <mihgen> amaksimov: who will be changing upgrade script? 16:32:07 <amaksimov> this is additional task which we didn't plan 16:32:13 <amaksimov> that's why I am against it 16:32:14 <mihgen> do we have patchsets on review for it? 16:32:21 <amaksimov> nope 16:32:36 <mihgen> do we have anyone here who is working on it? 16:32:36 <amaksimov> this is not just bugfix 16:32:39 <amaksimov> this is feature 16:32:49 <mihgen> this is feature; I agree 16:32:59 <amaksimov> so I was against it 16:33:06 <amaksimov> because it is too late 16:33:12 <amaksimov> for such big changes 16:33:25 <mihgen> angdraug: do you know if vkozhukalov/dpyzhov was consulting about upgrade scripts 16:33:27 <amaksimov> and the fact that we are removing resource from fuel 7.0 16:33:31 <angdraug> amaksimov: mihgen: please invite people actually working on centos6.6 if you want to have a meaningful discussion of it here 16:33:41 <mattymo_> why can't we just tell the user to reboot after? 16:34:40 <amaksimov> well, we can.. but again, we need to think about UX etc 16:35:08 <amaksimov> this should be tracked as feature. not just like an ordinary bug 16:35:09 <angdraug> good form would have been to add it to agenda yesterday and invite the feature lead 16:35:10 <mihgen> ok guys we can't discuss anything here since relevant people are not here. Let's follow up on this 16:35:38 <mihgen> angdraug: good form would be to bring yourself to the meeting if you are hacking features in fuel 16:35:51 <angdraug> amnk: please comment on centos 6.6 impact on upgrade script? 16:36:15 <amnk> angdraug: it has no impact to my knowledge 16:36:43 <angdraug> amaksimov: what makes you think there will be an impact? 16:36:45 <amaksimov> how we can upgrade kernel w/o reboot amnk ? 16:36:49 <mihgen> amnk: could you please update your name in IRC settings? 16:36:49 <evgenyl> I'm not sure if upgrade script should restart the nodes, it's dangerous action and we agreed that it should be user's conscientious decision. 16:37:04 <angdraug> evgenyl: +1 16:37:10 <mihgen> evgenyl: +1 16:37:25 <angdraug> amnk is Aleksandr Mogylchenko, feature lead for centos 6.6 16:37:40 <amnk> amaksimov: well, it is your decision to reboot node or not. It will work in any case 16:38:15 <angdraug> if we are not going to amend upgrade script to reboot the node, and I agree that we shouldn't, what other impact is there? 16:38:26 <amaksimov> amnk but you will have to notify user about it 16:38:32 <amnk> amaksimov: why? 16:38:37 <mattymo_> evgenyl, +1 to user initiated reboot 16:39:46 <ogelbukh> currently upgrade script tell user to run 'yum update; docker destroy all; docker start all' 16:40:17 <mihgen> angdraug: as you've been in the discussion, do we have committment to be on the hook to fix bugs / support upgrade procedure if needed? 16:40:17 <ogelbukh> adding 'reboot' here doesn't seem to add too much manual action :) 16:40:35 <angdraug> amnk has committed to that, yes 16:40:51 <mihgen> ok that's good 16:40:56 <angdraug> amnk: right? 16:41:03 <amnk> yes 16:41:06 <mihgen> amnk: did you guys check scale up story after upgrade? 16:41:30 <mihgen> let's say you have env with openstack, then you upgrade fuel, and now you want to add couple of nodes to old env 16:41:56 <mihgen> I'm about centos-based env specifically 16:42:24 <amnk> mihgen: this is for 7.0 - there are no centos-based envs there. 16:42:49 <angdraug> amnk: mihgen is talking about a 6.1 env deployed by a 6.1 fuel before upgrade to 7.0 16:44:09 <amnk> as I said in the email already - I could not get upgrade feature to work on stock 6.1. And others confirmed that it is not stable. So I do not have reliable test case to rely on 16:44:41 <mihgen> upgrade of Fuel? 16:44:49 <amnk> yes. 16:44:54 <mihgen> nurla: ^^ ? 16:45:19 <mihgen> can you please comment on stability of upgrade procedure? I've not heard about major issues for 6.0 -> 6.1 16:45:39 <mihgen> regardless, it doesn't mean that we just drop this from our testing matrix 16:45:45 <mihgen> and declare that we don't support it 16:45:46 <angdraug> has anyone tested 6.1-7.0 yet? 16:46:11 <mihgen> if we are about to deprecate a feature, it has to be done properly 16:46:18 <amnk> I just said that upgrade feature can not be used as reliable test case. 16:46:31 <amnk> I'm not saying we should deprecate it 16:46:40 <angdraug> amnk: doesn't have to be reliable, you only have to prove that you can get it to work 16:46:49 <mihgen> this is not a test case itself. angdraug +1 16:47:03 <mihgen> it 16:47:30 <mihgen> user use case is simple, I had 6.1 with centos; now I've heard 7.0 is cool and has kilo support 16:47:42 <mihgen> so I upgrade my fuel keeping old env on centos 16:47:53 <mihgen> and now I try to scale up old centos-based env 16:48:03 <mihgen> or do any other operations on that one 16:48:42 <amnk> well, if that user is able to get it working there will be no problems after we merge our custom build of kernel. 16:48:46 <mihgen> If we can't make it as such, then we would better not to even recommend upgrade to 7.0, and instead provide manual precedure 16:49:37 <angdraug> amnk: that's exactly what mihgen wants you to confirm 16:49:50 <amnk> angdraug: I confirm 16:50:20 <mihgen> amnk: the problem can be in cobbler snippets for example 16:50:53 <mihgen> if you update any of those as part of 6.6 work, ensure that it stays backward compatible 16:51:07 <amnk> mihgen: if there is a problem - I'll fix it. 16:52:08 <angdraug> I'm concerned that no one answered my question about testing 6.1-7.0 upgrade 16:52:27 <bookwar> "I could not get upgrade feature to work on stock 6.1" - we have upgrade tests which test ubuntu deployments after 6.1 to 7.0 upgrade, and they are green. Afaik Centos-based are work in progress by qa team 16:52:45 <angdraug> thanks 16:53:33 <angdraug> nurla: can you update on status of testing centos envs after upgrade to 7.0? 16:54:21 <angdraug> amnk: if the auto test of upgrades to 7.0 is green, you should be able to test it manually 16:54:59 <angdraug> we're running out of time, lets sum up 16:55:23 <angdraug> 1) there's no impact on Fuel, including upgrade script, but its message should be updated to recommend Fuel node reboot 16:56:08 <angdraug> 2) upgrades from 6.1 to 7.0 are verified by swarm and are green, so amnk is not blocked with testing 6.1 centos env scale up post-upgrade 16:56:24 <angdraug> any other blockers? 16:57:05 <amaksimov> just common sense - big change after FF 16:57:09 <sgolovatiuk> swarm doesn't check customized environments or environments with plugins 16:58:06 <mihgen> this is a big change, and risks have to be taken seriously 16:58:10 <amnk> please stop calling it a "big change". It is not. 16:58:23 <bookwar> sgolovatiuk: that's generic issue, for any kind of upgrade 16:58:38 <sgolovatiuk> amnk: it's a big change 16:58:41 <mihgen> yes it is - there are many things which can be affected 16:58:58 <angdraug> we already have centos 6.6 in the fuel docker containers 16:59:03 <mihgen> and we don't have any thorough test matrix for it 16:59:04 <amnk> sgolovatiuk: it was done silently inside docker containers, and nobody even noticed. 16:59:40 <mihgen> amnk: no one would notice change centos -> debian in containers 16:59:41 <mihgen> but not here 16:59:50 <angdraug> biggest part of it all is kernel upgrade 16:59:52 <sgolovatiuk> Do we test such scenarios? Do we test upgrades on production (live) environments? Do we have any --noop run? 16:59:53 <mihgen> when you change bootstrap as well, and other things 17:00:04 <mihgen> are affected, such as upgrade of fuel 17:00:14 <tmcpeak> o/ 17:00:16 <hyakuhei> o/ 17:00:18 <angdraug> but if a kernel upgrade to a different patch level within the same major version is a feature for us we have a bigger problem 17:00:23 <angdraug> time folks 17:00:28 <mihgen> xarses: closing 17:00:36 <IvanKliuk> bye! thanks! 17:00:49 <xarses> #endmeeting