Wednesday, 2023-10-04

opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: Generate hostnqn  https://review.opendev.org/c/openstack/kolla-ansible/+/89722706:33
opendevreviewPierre Riteau proposed openstack/kayobe master: Revert "CI: Disable bare metal testing on RL9/c9s"  https://review.opendev.org/c/openstack/kayobe/+/89726006:55
opendevreviewPierre Riteau proposed openstack/kayobe master: Revert "CI: Disable bare metal testing on RL9/c9s"  https://review.opendev.org/c/openstack/kayobe/+/89726006:57
mnasiadkamorning07:19
SvenKieskemorning07:24
SvenKieskeI need some vacation after my vacation :D07:25
mnasiadkanothing new07:28
opendevreviewBartosz Bezak proposed openstack/kolla-ansible master: Default keystone user role changed to member  https://review.opendev.org/c/openstack/kolla-ansible/+/89722207:30
mnasiadkaSvenKieske: welcome back ;)07:33
SvenKieskethanks :)07:35
SvenKieskethat's a nice feature addition: https://review.opendev.org/c/openstack/kolla-ansible/+/89722707:36
mnasiadkawell, that's a bug I would say :)07:37
SvenKieskealright, so it could be a bug worth a "fixes" relnote then? ;)07:38
rockeyisn't it easier to utilize copy module with content, instead of a one-row template?07:49
rockeyi mean, the complexity of that template isn't really there :)07:49
opendevreviewPierre Riteau proposed openstack/kayobe stable/2023.1: Fix data file path detection with new pip  https://review.opendev.org/c/openstack/kayobe/+/89637907:56
opendevreviewPierre Riteau proposed openstack/kayobe stable/zed: Fix data file path detection with new pip  https://review.opendev.org/c/openstack/kayobe/+/89638007:59
opendevreviewPierre Riteau proposed openstack/kayobe stable/yoga: Fix data file path detection with new pip  https://review.opendev.org/c/openstack/kayobe/+/89638107:59
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: Generate hostnqn  https://review.opendev.org/c/openstack/kolla-ansible/+/89722708:15
SvenKieskemnasiadka: didn't we already have a planning etherpad for ptg? I can't find it at https://etherpad.opendev.org/p/KollaWhiteBoard08:48
mnasiadkahttps://etherpad.opendev.org/p/kolla-caracal-ptg (https://ptg.opendev.org/etherpads.html)08:49
SvenKieskeadded it to the list on the WhiteBoard08:50
mnasiadkathanks08:50
SvenKieskeshould probably add some content for the PTG as well; I have no prior experience how much we will be able to discuss, to me it seems a little empty?08:52
hrwSvenKieske: check previous ones for examples?09:00
SvenKieskehrw: that's a good idea!09:04
SvenKieskewill try to do that later, currently swamped in meetings :)09:04
opendevreviewPierre Riteau proposed openstack/kayobe master: Use importlib.metadata instead of importlib_metadata  https://review.opendev.org/c/openstack/kayobe/+/89730609:13
opendevreviewPedro Henrique Pereira Martins proposed openstack/kolla-ansible master: Fix the docker container dimensions comparison for short notation  https://review.opendev.org/c/openstack/kolla-ansible/+/88650009:26
opendevreviewCélestin Matte proposed openstack/kolla-ansible master: Ansible-lint: crash when warnings occur  https://review.opendev.org/c/openstack/kolla-ansible/+/89731210:03
SvenKieskedid we already enable rabbitmq quorum queues? I lost track..10:14
opendevreviewPierre Riteau proposed openstack/kayobe stable/2023.1: Use importlib.metadata instead of importlib_metadata  https://review.opendev.org/c/openstack/kayobe/+/89731310:20
opendevreviewPierre Riteau proposed openstack/kayobe stable/zed: Use importlib.metadata instead of importlib_metadata  https://review.opendev.org/c/openstack/kayobe/+/89731410:20
kevkomnasiadka: i jus tbuilt images wih kolla and i am missing ironic-dnsmasq 10:23
mnasiadkakevko: why?10:24
kevkowhat i am doing wrong ... do we miss it ? 10:24
kevkoaaaa, it is different name ..it's just dnsmasq ..not ironic-dnsmasq10:29
mnasiadkahaha10:31
opendevreviewMichal Nasiadka proposed openstack/kolla-ansible master: Generate hostnqn  https://review.opendev.org/c/openstack/kolla-ansible/+/89722710:59
opendevreviewBartosz Bezak proposed openstack/kolla-ansible master: Default keystone user role changed to member  https://review.opendev.org/c/openstack/kolla-ansible/+/89722211:07
mnasiadkamgoddard mnasiadka bbezak frickler kevko SvenKieske mmalchuk gkoper jangutter jsuazo - meeting in 712:53
mnasiadka#startmeeting koll13:00
opendevmeetMeeting started Wed Oct  4 13:00:54 2023 UTC and is due to finish in 60 minutes.  The chair is mnasiadka. Information about MeetBot at http://wiki.debian.org/MeetBot.13:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.13:00
opendevmeetThe meeting name has been set to 'koll'13:00
mnasiadka#endmeeting13:00
opendevmeetMeeting ended Wed Oct  4 13:00:58 2023 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)13:00
opendevmeetMinutes:        https://meetings.opendev.org/meetings/koll/2023/koll.2023-10-04-13.00.html13:00
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/koll/2023/koll.2023-10-04-13.00.txt13:00
opendevmeetLog:            https://meetings.opendev.org/meetings/koll/2023/koll.2023-10-04-13.00.log.html13:00
mnasiadka#startmeeting kolla13:01
opendevmeetMeeting started Wed Oct  4 13:01:02 2023 UTC and is due to finish in 60 minutes.  The chair is mnasiadka. Information about MeetBot at http://wiki.debian.org/MeetBot.13:01
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.13:01
opendevmeetThe meeting name has been set to 'kolla'13:01
mnasiadka#topic rollcall13:01
mnasiadkao/13:01
SvenKieskeo/13:01
janguttero/13:01
bbezako/13:01
mmalchuko/13:01
frickler\o13:02
mhinero/13:02
mnasiadka#topic agenda13:03
mnasiadka* CI status13:03
mnasiadka* Release tasks13:03
mnasiadka* Regular stable releases (first meeting in a month)13:03
mnasiadka* Current cycle planning13:03
mnasiadka* Additional agenda (from whiteboard)13:04
mnasiadka* Open discussion13:04
mnasiadka#topic CI status13:04
mnasiadkaAfter RMQ drama I think we're back into business as usual13:04
mnasiadkaThere have been some post failures on pushing to quay.io13:05
mnasiadkabut let's monitor that13:05
mnasiadka#topic Release tasks13:06
mnasiadkaI think none really, we're waiting for RDO to publish stable/2023.213:06
mnasiadkato switch to those repos13:07
mnasiadkanothing apart this13:07
mnasiadka#topic Current cycle planning13:07
mnasiadkaSo, I didn't have time looking into LE or Podman13:07
mnasiadkaBut I'll try to make kevko happier by doing it this week13:07
mnasiadkaanybody else having cycles to look into those two?13:08
SvenKieskeI'll try to, but probably later this week, at least not today I think, still other stuff to do13:08
mnasiadka#link https://review.opendev.org/c/openstack/kolla/+/88734713:08
mnasiadkaand 13:09
mnasiadka#link https://review.opendev.org/c/openstack/kolla-ansible/+/74134013:09
mnasiadkakevko: the k-a patch has a merge conflict - please update :)13:09
mnasiadkaSvenKieske: how are we with td-agent upgrade? need help?13:09
SvenKieskeI'd hope to finish that, I'm not aware of "more" issues then the currently known ones: file paths, user changes13:10
opendevreviewJuan Pablo Suazo proposed openstack/kolla master: Adds TAAS Neutron plugin to support OVS port mirrors  https://review.opendev.org/c/openstack/kolla/+/88515113:10
SvenKieskeI guess I asked last week about whether we care about UID|GID stability and never received an answer13:11
SvenKieskeI know there is some code which pins users to certain UIDs I can't imagine a way how this could work with the new fluent package13:12
SvenKieskeso in that area, if we care about it, help would be nice, I guess it might be configurable in the new fluentd-package, but I didn't have time to trace it down just yet13:12
mnasiadkado we need to change the user?13:12
fricklerwell we need some way of upgrading existing things I guess? pinning would be easier than having a migration script13:13
SvenKieskeyes, the new username is "_fluentd" if my memory is correct and this is going to be set during "postinst" phase during package install13:13
mnasiadkafantastic13:13
SvenKieskeI did already mention this on IRC and in the changeset discussion with no feedback so far :)13:13
mnasiadkawell then we need to add that user with new UID13:14
mnasiadkadoes that mean we need to change some permissions on upgrade? uhh13:14
SvenKieskemhm13:14
SvenKieskenot sure what is stored in these paths tbh13:14
mnasiadkado we have any other choice? I don't think so13:14
SvenKieskequestion is what would be lost if we have /var/lib/fluentd/ instead of old path (don't nail me down on it, the important bit is /var/lib/$bla changes)13:15
mnasiadkaprobably nothing, but if you don't wander this route - you won't find out I guess13:16
SvenKieskeactually don't know what fluentd stores there, beside in flight data, sockets etc13:16
SvenKieskeI'll check it out13:16
mnasiadkawe should be fine, some buffers, in flight data, sockets13:19
mnasiadkaI've also been working on https://review.opendev.org/c/openstack/kolla-ansible/+/897227 - it's a bugfix, but given the amounts of comments in the bug it seems to be important13:20
mnasiadkaSeems people are using nvmeof after all13:20
mnasiadka#topic Additional agenda (from whiteboard)13:21
mnasiadkalet's go through that, there's some13:21
mnasiadka(jsuazo) 844614: Deploy Glance and Cinder Backup with S3 backend support 13:21
mnasiadka#link https://review.opendev.org/c/openstack/kolla-ansible/+/84461413:21
jsuazoThat proposal has addressed all comments and should be ready to merge :D13:22
mmalchukconfirm that13:22
mnasiadkaWhat about https://review.opendev.org/c/openstack/kolla-ansible/+/844998/1 ?13:23
mnasiadkaShould I abandon this?13:23
mmalchukyep13:24
jsuazoYes, those changes are included in the proposal13:24
mnasiadkaI still don't know if I like the s3_* variables13:25
mnasiadkafrickler: willing to have a review?13:25
mmalchuknorm variables)13:25
SvenKieskeare they to short? :D13:25
mnasiadkawell, usually we should have role specific variables13:26
SvenKieskejust add it to the review and discuss there? :)13:26
mnasiadkabut if we want to make it easier for users13:26
mnasiadkaI guess it's ok13:26
kevko\o/ 13:26
SvenKieskecould make sense13:26
kevkosorry for delay 13:26
mnasiadkano worries kevko13:26
mnasiadkanext one13:27
mnasiadka(jsuazo) 885417: Configures the tap-as-a-service neutron plugin13:27
SvenKieskecould also be more complicated for the users if they want different s3 URLs/buckets :)13:27
mmalchukthe review is an year old, lets merge it13:27
mnasiadka#link https://review.opendev.org/c/openstack/kolla-ansible/+/88541713:27
mnasiadkaSvenKieske: it's covered in the docs13:27
SvenKieskeok :)13:27
mnasiadkajsuazo: did we merge the Kolla side?13:28
mnasiadkaAh no, we didn't13:28
jsuazonot yet. its the next point on the list13:28
mnasiadka#link https://review.opendev.org/c/openstack/kolla/+/88515113:28
mmalchuknot yet13:28
jsuazoZuul failing on unrelated image builds, rebased again and waiting for the jobs to run13:28
mnasiadkaok then13:29
jsuazoBut made the changes to make the installation simply through pip13:29
mmalchukwe change from source to pypi and I'm not sure13:29
fricklerplease don't rebase if there is no merge conflict13:29
* frickler feels like a broken record repeating that13:29
mmalchukit work perfect from source13:29
mnasiadkareviewed13:31
jsuazoIt did, but as the source is not very well maintained (branch,version and tag-wise) its simpler to do it through pip13:31
kevkowhy you don't like rebase ? 13:31
kevkoif no merge conflict 13:31
fricklerbecause it breaks patch history and is plain unneccessary13:31
mmalchukfrickler care about CI cpu's ;_13:31
fricklerCI can be triggered with a recheck comment just as well13:31
SvenKieskeI'm always a big fan of not doing unneccessary work :)13:31
fricklerand reviewers don't have to go looking what might have changed through umpteen unneeded rebases13:31
mmalchukfrickler we don't have history) with git amend))13:31
mnasiadkafrickler: we need a recheck button ;-)13:31
mmalchukbig and red )))))13:32
fricklerno, people need to think before they recheck and add a meaningful reason13:32
fricklerelse I'll complain about that, too ;-P13:32
mnasiadkahaha13:33
mnasiadkaok, let's move on13:33
SvenKieskethat's also true13:33
mnasiadka(jsuazo) 875983: Enable the Fluentd Plugin Systemd13:33
SvenKieskebut what about that pip bit now?13:33
mnasiadka#link https://review.opendev.org/c/openstack/kolla-ansible/+/87598313:33
mnasiadkaSvenKieske: nothing, should work13:33
SvenKieskeok13:33
SvenKieskethe systemd plugin is waiting on your feedback, I did propose to make this dependent on "enable_central_logging" which defaults to "no"13:34
SvenKieskeah no, I see the feedback was already addressed.13:35
jsuazoAdded a variable to opt in to the fluentd systemd, "enable_fluentd_systemd",it defaults to the value of "enable_central_logging"13:35
mnasiadkareviewed, docs oneliner and reno fixup needed and looks fine13:36
SvenKieskethat sentence should go into the release notes imho13:36
SvenKieskeand|or in the docs13:36
SvenKieskenot sure how we cover every variable, I guess we don'13:36
SvenKiesket have everything in the docs..13:36
mnasiadkawe don't, but let's try to make docs better, not worse :)13:36
jsuazoagree, i'll add it while addressing the new coments13:36
SvenKieskenice, ty13:37
mnasiadka#topic Open discussion13:37
mnasiadkathanks13:37
mnasiadkaso - anybody anything?13:37
mnasiadkaAh right13:37
mnasiadkaI have one13:37
mnasiadkaPTG timing13:37
mnasiadkafrickler: you didn't vote on the poll - so can you state your preferences here please?13:38
fricklerI don't care, open to anything13:39
SvenKieskeyeah, good news: our downstream meetings are mostly canceled so I guess I should've (more) time for PTG :)13:39
mnasiadkaok, seems everybody is open to anything13:39
SvenKieskeI did vote :)13:39
mnasiadkausually we did a very late slot to allow US to join13:39
mmalchukme too13:39
mnasiadkabut I haven't seen anybody from US since long time13:40
mmalchukanybody is here?)13:40
jangutterDoes that mean we can do s/color/colour/g finally?13:40
kevkoi have something 13:40
mnasiadkaseems not13:41
mnasiadkaBobcat PTG was 13:00 - 17:00 UTC13:41
mnasiadkaso 15:00 - 19:00 CEST13:41
mnasiadkaare we sticking to that, or should we shift a bit earlier?13:41
fricklerif you are asking it that way, I'd say earlier13:42
mmalchuk12:59 UTC ?13:42
SvenKieskeone or two hours earlier would be nice I guess.13:42
mmalchukbe more specific13:43
SvenKieske12:00 - 16:00 UTC ?13:43
mmalchukone hour is ok 13:43
mnasiadkaok, let's shift one hour13:43
mnasiadkait's still east coast US friendly13:43
fricklerjust note that that won't appear in the official timetable13:44
mmalchukstill nobody from US13:44
fricklerbut the TC has been being similar things, so it should be ok13:44
SvenKieskemhm, is it to late to change the official timetable?13:44
frickleryes13:45
SvenKieskethen we should at least send out an email etc. to notify ppl not living on IRC everyday ;)13:45
kevkoGuys, I need to go out of PC for 10 minutes ..but I would like to also ask several things, so let me ask now and we can discuss after several minutes ...13:45
SvenKieskeok13:46
mnasiadka23-27 October 2023:13:46
mnasiadkaMonday - 12.00 - 15.00 UTC (general, Kolla and Kolla-Ansible)13:46
mnasiadkaTuesday - 12.00 - 14.00 UTC (Kolla-Ansible)13:46
mnasiadkaTuesday - 14.00 - 15.00 UTC (Operator Hour)13:46
mnasiadkaThursday - 12.00 - 14.00 UTC (Kayobe)13:46
mnasiadkaIs that basically what we would like?13:46
SvenKieskeseems fine to me13:46
mnasiadka3 hours on Monday, 2+1 hours on Tuesday and Kayobe on Thursday13:46
frickler+113:46
mmalchukfine13:46
SvenKieskenot 12-16?13:47
SvenKieskeone hour less?13:47
fricklerwith no topics to discuss, we'll be even faster13:47
SvenKieskeI don't know, just asking, my first PTG13:47
mnasiadkawell, last time we didn't have that many topics, we can do 12-16 and just finish earlier13:47
kevko1. Rabbitmq/erlang incompatibility ... We found incompatibility in erlang/rabbitmq in yoga.. For example now I am checking master rabbitmq ...paste here -> 13:47
kevkohttps://paste.openstack.org/show/b2voIvMhqW4IblzVE5jD/13:47
mnasiadka3.12 in Yoga?13:48
kevkono ..13:48
kevkothat paste is for master 13:48
mmalchukkevko may be move to etherpad ?13:48
kevkobecause currently i've decided to check all debian images from master -> antelope -> ...13:49
mnasiadkakevko: can you post a summary instead of shredded bits?13:49
kevkookay i  will 13:49
SvenKieskemhm, but shouldn't master be on erlang vm 26 already?13:49
mnasiadkathat would be easier to understand and fix :)13:49
mnasiadkaprobably should be, and kevko is only checking Debian13:49
mnasiadkacan we get a full view on all distros and then think if we need to fix anything?13:49
kevkoyeah, but question was that i think we should switch erlang/rabbit to novemberrain ..and pin every version in every branch to avoid this in future13:49
mnasiadkarabbit is from novemberain13:50
kevkoand erlang should be also i think 13:50
mnasiadkaerlang can't be switched, because they don't mirror aarch6413:50
kevkospecial case is arm13:50
mnasiadkaand they use launchpad to build the packages13:50
mnasiadkaso why would we download from a mirror that downloads from launchpad?13:50
kevkobut arm has different config in kolla13:50
mnasiadkawhat do you mean by different?13:51
kevkorepos.yaml13:52
mnasiadkaI don't understand, please describe the situation fully somewhere - so we can all analyse13:53
kevkoSecond thing -  there is no option to configure logging in kolla-ansible .. (i mean, formatters, handlers ...etc ....) ..so i will probably propose a new shared role 13:53
kevkookay i will 13:53
kevkoi will add to etherpad probably 13:54
mnasiadkakevko: that's a PTG topic most probably13:54
mmalchukbig plans13:54
kevkoand last question ..is it ok to add repo for python3-podman for ansible collection which is installing this package  ? 13:54
mnasiadkasure13:54
kevkobecause i will need to create a ticket for debian release team and kindly ask them to include my package into stable 13:55
kevkofor now it is in unstable ..13:55
mnasiadkaok, coming back to the topic13:55
mnasiadka3 hours times two for the PTG is enough, or do we prefer 4?13:55
mnasiadka(I'm fine with both)13:55
mmalchuk3 hours for rabbitmq fair enough13:55
mnasiadkaFor RabbitMQ 3 weeks are not enough13:56
mmalchukdo we have more topics?)13:56
* SvenKieske actually debugged some rabbit stuff earlier this day13:56
SvenKieskeI asked earlier today about the status of quorum queues13:57
mnasiadkaok, let's go for 3 hours on Mon and Tue, I'm pretty sure you'll all be sleeping on those sessions anyway :)13:57
mmalchuksure13:57
SvenKieskeI did only find out it's apparently support in oslo now, but I think we didn't implement it yet in kolla?13:57
mnasiadkaSvenKieske: not yet in Kolla, feel free to propose ;)13:57
mnasiadkaok then13:58
SvenKieskemight be a topic for PTG? or is it a no-brainer? :D13:58
mnasiadkaprobably more implementation detail than topic for PTG, we could discuss if that should be default or not :)13:58
mnasiadkabut sure, add it to the list13:58
SvenKieskeokay then13:58
mnasiadkathe more topics we have, the less chance we finish in 7 minutes13:58
mnasiadkakevko: does the PTG timeline suit you?13:59
SvenKieskePTG board, btw:13:59
SvenKieske#link https://etherpad.opendev.org/p/kolla-caracal-ptg13:59
mnasiadkathanks SvenKieske 13:59
mnasiadkaok, time is up13:59
mnasiadkathanks for coming :)13:59
mnasiadka#endmeeting13:59
opendevmeetMeeting ended Wed Oct  4 13:59:49 2023 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)13:59
opendevmeetMinutes:        https://meetings.opendev.org/meetings/kolla/2023/kolla.2023-10-04-13.01.html13:59
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/kolla/2023/kolla.2023-10-04-13.01.txt13:59
opendevmeetLog:            https://meetings.opendev.org/meetings/kolla/2023/kolla.2023-10-04-13.01.log.html13:59
mmalchukthanks mnasiadka 13:59
SvenKieskethanks14:01
mnasiadkaof course ptg bot does not have the 12.00 UTC slot14:13
greatgatsbysorry I'm reading back in the logs here and trying to understand the rabbitmq issue.  We're on kolla-ansible yoga and upgraded our docker images/containers yesterday and we're seeing all kinds of rabbitmq "crashed queue" type messages.  Is this related?  Is there a ticket somewhere we can follow what's happening?14:14
mnasiadkaI would suggest filing a ticket in Launchpad for Kolla14:16
greatgatsbykevko mentioned a "incompatibility in erlang/rabbitmq in yoga" ... just trying to understand if there's a known issue with rabbitmq and yoga right now?14:17
mnasiadkaSvenKieske, frickler: that probably means we need to switch to 13:00 UTC or work against the official PTG schedule14:29
fricklerwell that's what I was trying to imply earlier. but who cares about the ptg bot? just add the timeframe to our etherpad and be done14:37
mnasiadkayeah, added that14:38
opendevreviewMerged openstack/kolla-ansible master: Default keystone user role changed to member  https://review.opendev.org/c/openstack/kolla-ansible/+/89722214:58
opendevreviewBartosz Bezak proposed openstack/kolla-ansible stable/2023.1: Default keystone user role changed to member  https://review.opendev.org/c/openstack/kolla-ansible/+/89728714:58
opendevreviewBartosz Bezak proposed openstack/kolla-ansible stable/zed: Default keystone user role changed to member  https://review.opendev.org/c/openstack/kolla-ansible/+/89728814:58
opendevreviewBartosz Bezak proposed openstack/kolla-ansible stable/yoga: Default keystone user role changed to member  https://review.opendev.org/c/openstack/kolla-ansible/+/89728914:59
SvenKieskewell I'd care about it, because it should be discoverable to external people when we meet, so they can participate, no?15:09
fricklerthey can read the agenda and discover it, if they don't do that, they can still join an hour late IMO15:25
kevkogreatgatsby: yeah, there is different unsupported version of erlang :D 15:32
greatgatsbykevko: thanks for the reply, sorry for the poorly worded questions.  We had a number of VMs crash/power down last night.  We're seeing a lot of rabbit errors (no surprise there) but we're also seeing "Instance shutdown by itself" from nova logs during a kolla-ansible deploy.15:41
greatgatsbythe deploy was just upgrading the kolla containers, which would have restarted most services15:42
kevkogreatgatsby: that's exactly the same error i've seen before i fixed image 15:48
greatgatsbywhat fix?  Could you please point me to a ticket or something with more info?  Greatly appreciated!15:49
kevkogreatgatsby: well, i still didn't send to upstream ...but i just pinned erlang version to proper version15:51
kevkogreatgatsby: currently i am working on a fix15:51
kevkogreatgatsby: fix is very easy ...but i would like to provide some script which will check it automatically and fail in case it's bad 15:51
greatgatsbyI see, so this is due to the rabbitmq/erlang versions in the rabbitmq image?  I'm seeing rabbitmq-server 3.9.29-1 and erlang 1:25.3.2.6-1 in ours.15:56
greatgatsbykevko: thanks again for the help, never would have discovered this on my own.  Are you saying I need to poing erlange to 25.2?  That seems to be what the rabbitmq docs show as the max erlang version?16:04
kevkogreatgatsby: https://www.rabbitmq.com/which-erlang.html16:24
greatgatsbykevko: excellent, thanks.  That's where I got the 25.2 version from (for 3.9.29) which does seem to indicate the 25.3 in the container is a version mismatch?  Just want to make sure I have my facts straight.16:25
kevkogreatgatsby: yep16:26
greatgatsbythanks!16:27
mnasiadkakevko: so from where do we take erlang 25.2 in Yoga? ;-)16:29
mnasiadkaor we pin to 2416:29
kevkoi just created a pin to 25.2 :D 16:30
kevkonovemberrain ? 16:30
mnasiadkano aarch6416:30
mnasiadkalaunchpad only has latest build16:30
mnasiadkaso that leaves us with 2416:30
mnasiadkawhat concerns me a bit, is why the heck CI works :D16:31
kevkomnasiadka: so why we have ./kolla/template/repos.yaml ??? 16:31
kevkothere are separate blocks for aarch64 ? 16:31
mnasiadkakevko: And you want to have different versions for x86 and aarch64?16:32
mnasiadkaWhile we strived for cycles to have the same versions on every distro and arch?16:32
kevkomnasiadka: another option is to have it non-working :) 16:32
mnasiadkaerlang 24 is supported with rmq 3.916:33
mnasiadkaso what's the problem? :)16:33
kevkoCI is working ..and actually ... openstack is working ...but it is crashing from time to time ...and it's definitely rabbitmq/erlang incompatibility 16:33
kevkoi am not expert for rabbitmq/erlang 16:33
kevkobut what if you run upgrade from erlang 25 -> 24 ? 16:33
kevkoi would say that it won't be without problems :) 16:34
mnasiadkaNobody is an expert, nobody knows :)16:34
kevkoi am 90 percent sure that it will be problem :D 16:35
mnasiadkawell, there's 25.3 already16:35
mnasiadkaso either way, we will downgrade16:36
mnasiadka;)16:36
mnasiadkahmm, the same is on centos/rocky16:36
mnasiadkafantastic16:36
kevkonow i don't understand 16:37
mnasiadkawe published 25.3 already16:37
mnasiadkaso if somebody moved to 25.316:37
mnasiadkaand now he gets 25.216:37
mnasiadkaisn't that a downgrade?16:37
kevkoi think .2 is less than .3 no ? :D 16:38
mnasiadkaand centos already has erlang-25.3.2.6-1.el8.x86_6416:38
mnasiadkaand nobody complained for months16:38
kevkoyeah, that's the issue we are talking about :D :D 16:38
kevkoi was complaining and greatgatsby also :D 16:39
kevkoand documentation saying "maximum supported" ...16:39
kevkoerlang/rabbitmq is very sensitive 16:39
mnasiadkawell, it's been like that for long16:39
mnasiadkaI'm pretty sure erlang 25.3 didn't show up even this year16:39
mnasiadkawell, 25.3 was released in March16:40
mnasiadkathis year16:40
mnasiadkaso probably anybody that did a deploy on a new container since march - should be observing that16:41
mnasiadkaquestion why there's no bugs raised :)16:41
kevkowhen you realize how often people upgrade the openstack release, and especially rabbitmq ... 6 months is not so long that it is possible that people are not aware of this error16:42
mnasiadkaprobably16:42
mnasiadkaI'll run a test tomorrow what happens when we downgrade ;)16:42
kevkowell, because i haven't reported yet :D 16:42
mnasiadkaI hope downgrading from 25.3 to 25.2 won't blow out your cluster16:42
kevkowe found it on new deployment on customer side 16:42
kevkoi've already done it 16:43
kevkoand it's fixed ..16:43
mnasiadkawhere is it fixed?16:43
kevkono crash, no turn off the rabbitmq ..nothing ..working like a charm several days from fix16:43
kevkomy downstream repo and my image ...16:43
mnasiadkayes, but that might not be the fix for everybody, unless we drop aarch64 :)16:44
kevkobecause you know ..sometimes fix in upstream can take days ...and you have to convince many people that you are right :D 16:44
kevkoand i am : 16:44
kevko:) 16:44
kevkowell, i think we already have different situation with aarch6416:45
mnasiadkathe version is the same there16:45
kevkoi mean it's is somehow specific 16:45
mnasiadkaI know that probably nobody is using that :)16:45
mnasiadkaYou can say that out loud16:46
kevkoso why to not focus to debian/ubuntu centos/rocky ...... to have same compatible versions and working ..and aarch64 also even if they has different version16:46
kevkoi know that it is not ideal situation ...but ...16:46
kevko:D 16:46
kevkohaha 16:46
kevkoyeah , probably nobody :D 16:46
kevko^^^^^ then you can use jinja apt and pin to supported versions for both archs but different versions 24/25 for erlang for example 16:47
mnasiadkajust writing the patch16:48
kevkoi also :D 16:48
kevkoAND ...i am working on script which will check every time the compatibility 16:48
kevkowe will just need to update supported matrix 16:48
kevkoand build will fail/pass based on check 16:48
kevkowhat do you think about it ? 16:49
mnasiadkaactually debian aarch64 uses erlang 2416:49
mnasiadkaah no16:49
mnasiadkaI missed that16:49
mnasiadkakevko: script? we could think of a framework for testing for correct versions of a lot of things in the container, but that's rather PTG topic16:51
mnasiadkawe just need to pin erlang correctly, which we have never done16:51
kevkoWell ..something as RUN check-compat ....which will compare some table of compatibility versions 16:52
kevkoI've pinned it in my downstream right after I figured out it is a problem 16:52
opendevreviewMichal Nasiadka proposed openstack/kolla stable/yoga: [yoga-only]: Fix erlang versions  https://review.opendev.org/c/openstack/kolla/+/89735216:53
kevkoWe had same errors as great Gatsby had16:53
mnasiadkaI haven't seen them on any deployment, and we have some fresh ones16:53
mnasiadkaBut better to stay on the safe side with EOL RMQ16:53
kevkoSo you don't believe it ? 🤣16:54
kevko*me16:54
kevkoWhy 23 ? 16:54
opendevreviewMichal Nasiadka proposed openstack/kolla stable/yoga: [yoga-only]: Fix erlang versions  https://review.opendev.org/c/openstack/kolla/+/89735216:54
mnasiadkatoo late16:54
mnasiadka;)16:54
mnasiadkakevko: I can't believe even fresh stable releases have EOL RMQ and we should do something about it16:55
opendevreviewMichal Nasiadka proposed openstack/kolla stable/yoga: [yoga-only]: Fix erlang versions  https://review.opendev.org/c/openstack/kolla/+/89735216:55
kevkoYep but that is another issue 16:56
mnasiadkaok, that should be enough16:56
mnasiadkaand let's try to solve this mess on the PTG16:58
mnasiadkakevko: you're going to join this time? ;)16:58
kevko28 leaving to vacation on scheyschellas for 3 weeks :D 17:00
kevkosorry - 18th 17:00
kevkoso :( 17:00
mnasiadkagreat timing :)17:01
mnasiadkabut at least try to populate some ideas on the PTG etherpad :)17:01
kevkoi can somehow connect via mobile *maybe17:01
mnasiadkaAnd have fun on vacation :)17:01
mnasiadkanaah, vacation is vacation17:01
kevkobut i will be on a boat during that 3 weeks ..almost all the time 17:01
kevkodo you remember my old patch for logging ? 17:02
kevkothat would be very nice to revive17:02
mnasiadkaIf you're going to convince us this time this is needed - sure :)17:02
mnasiadkaneed to finish for today, talk to you another day ;)17:02
kevkowell - we definetly should provide option to pass own logging.conf17:03
kevkobecause by default there are several loggers in openstack ...and most of them are warns ... that's ok ...but sometimes you want to reconfigure loggers ... 17:04
kevkoespecially when debugging ...17:04
kevkoand that's not possible right now 17:04
kevkoalso - we are using fluentd ..but instead of fluentlogger we are writing logs to disk and tailing by fluentd :D 17:05
kevkowhich is not good ..17:05
mnasiadkaPut that to the ptg etherpad please :)17:06
kevkomnasiadka: last thing -> your pin will not work i think ..because ppa has only 25.3.2 17:06
kevkowhile novemberrain has all version of erlang ..17:06
mnasiadkaMight be, we need separate apt prefs for aarch6417:07
mnasiadkaWill update later17:07
kevkothat's why I've been talking that we should use mirrors for both rabbit/erlang packages 17:07
kevko()[root@pixla /]# apt-cache policy erlang-base17:08
kevkoerlang-base:17:08
kevko  Installed: 1:25.3.2.6-1rmq1ppa1~ubuntu20.04.117:08
kevko  Candidate: 1:25.3.2.6-1rmq1ppa1~ubuntu20.04.117:08
kevko  Version table:17:08
kevko *** 1:25.3.2.6-1rmq1ppa1~ubuntu20.04.1 50017:08
kevko        500 https://ppa.launchpadcontent.net/rabbitmq/rabbitmq-erlang-25/ubuntu focal/main amd64 Packages17:08
kevko        100 /var/lib/dpkg/status17:08
kevko     1:24.2.1+dfsg-1~bpo11+1 50017:08
kevko        500 http://osbpo.debian.net/debian bullseye-zed-backports-nochange/main amd64 Packages17:08
kevko     1:23.2.6+dfsg-1+deb11u1 50017:08
kevko        500 http://deb.debian.org/debian bullseye/main amd64 Packages17:08
kevko()[root@pixla /]# apt download erlang-base17:08
mnasiadkaYoga is going EM and then EOL so a couple more months only17:08
kevkothis is not about yoga ..but this is about rabbitmq packages 17:09
kevkomost of the time it's simple to cherry-pick fix from upstream 17:09
kevkobut ppa repo for erlang 25 don't have all versions ..17:10
kevkomirror has 17:10
mnasiadkaBut only for x8617:10
kevkoyeah - but it's another issue 17:10
kevkoif we separate x86 and aarch64 ... we can control both versions 17:11
kevkoapt url / pin 17:11
mnasiadkaAnd I’m not keen to move other branches unless we need to17:11
kevkobut we need it i think :D 17:11
mnasiadkaWhy?17:12
kevkohmm, maybe i didn't understand ...you said move other branches ...what does it mean ? 17:12
mnasiadkaAs in move other branches to use erlang from novemberain17:13
kevkoi think it is broken for all branches currently 17:13
kevkoand your pin will not work ..because there is not version what you need 17:14
kevko(erlang)17:14
mnasiadkaBroken for all branches? How is that possible?17:16
mnasiadka3.10 supports 25.3, so that works17:17
kevkoi didn't check all ... i just wanted to say that we were using url for apt ..but we need to take into account  that ppa don't have all versions ..so you don't have a control 17:18
kevkoyou don't have a control which version will be installed ..17:19
kevkoand if you add pin ... it can fail because there is no version which will fit your pinned version 17:19
kevko(novemberrain is something different ..they have all ..so you can just pin it )17:19
mnasiadkaI just prefer it fails build than breaks somebody’s deployment17:27
kevkomnasiadka: yeah, agree ...but when it fail (beacuse of pin and not available version in ppa) ...you will just need to switch to novemberrain17:35
opendevreviewMichal Nasiadka proposed openstack/kolla stable/yoga: [yoga-only]: Fix erlang versions  https://review.opendev.org/c/openstack/kolla/+/89735218:42
opendevreviewMichal Nasiadka proposed openstack/kolla stable/yoga: [yoga-only]: Fix erlang versions  https://review.opendev.org/c/openstack/kolla/+/89735218:44
opendevreviewPierre Riteau proposed openstack/kayobe stable/yoga: Fix data file path detection with new pip  https://review.opendev.org/c/openstack/kayobe/+/89638120:16
opendevreviewPierre Riteau proposed openstack/kayobe stable/yoga: Fix data file path detection with new pip  https://review.opendev.org/c/openstack/kayobe/+/89638120:23
opendevreviewVerification of a change to openstack/kayobe stable/2023.1 failed: Fix data file path detection with new pip  https://review.opendev.org/c/openstack/kayobe/+/89637922:56
opendevreviewMerged openstack/kayobe stable/zed: Fix data file path detection with new pip  https://review.opendev.org/c/openstack/kayobe/+/89638023:23

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!