Wednesday, 2024-11-13

opendevreviewMerged openstack/kayobe master: Prevent accidental overriding of Ansible extensions  https://review.opendev.org/c/openstack/kayobe/+/91776201:04
*** fungi is now known as Guest924901:33
*** kinrui is now known as fungi01:41
FyrieHow do I modify the base image to accept root CA? I am trying to build the kolla images via `kayobe overcloud container image build` I think I have to override Jinja2 blocks?05:55
Fyrieneed to add company's cert06:00
opendevreviewMatt Crees proposed openstack/kolla-ansible master: Update user role assignments  https://review.opendev.org/c/openstack/kolla-ansible/+/93441409:26
opendevreviewMatt Crees proposed openstack/kolla-ansible master: Give ironic-inspector system scope `all`  https://review.opendev.org/c/openstack/kolla-ansible/+/93441509:26
opendevreviewPierre Riteau proposed openstack/kayobe master: Add loop control to custom DNF repos  https://review.opendev.org/c/openstack/kayobe/+/93376109:58
lansonHello guys, back at it again wwith https://review.opendev.org/c/openstack/kolla-ansible/+/934429 which I believe should be good to go, and https://review.opendev.org/c/openstack/kolla-ansible/+/934747 which was split out of the cyborg fix and should also be fairly straight forward. cheers 09:59
SvenKieskemnasiadka: replied, there are quite some errors in at least one multinode job, not sure if I get around to dig into them today.09:59
SvenKieske( regarding https://review.opendev.org/c/openstack/kolla/+/934749 )09:59
mnasiadkaSvenKieske: I won't, I'm off today and tomorrow, and I just posted that so we start thinking about this10:17
SvenKieskesure, no problem :) thanks for doing so!10:21
opendevreviewRafal Lewandowski proposed openstack/kolla-ansible master: Enable ML2/OVN and distributed FIP by default.  https://review.opendev.org/c/openstack/kolla-ansible/+/90495911:18
opendevreviewMickael Razzouk proposed openstack/kolla master: Added libvirt dependency to nova-libvirt container  https://review.opendev.org/c/openstack/kolla/+/93495211:35
mootHello after upgrading my cluster to 2024.2 all mdev management broke, I submited a fix and wanted to ask if adding the mdevctl package here was the right way of doing it 11:38
opendevreviewMerged openstack/kayobe stable/2024.1: Add internal VIP address to no_proxy  https://review.opendev.org/c/openstack/kayobe/+/93472911:46
opendevreviewMerged openstack/kayobe stable/2023.2: Add internal VIP address to no_proxy  https://review.opendev.org/c/openstack/kayobe/+/93473012:03
opendevreviewMerged openstack/kayobe stable/2023.1: Add internal VIP address to no_proxy  https://review.opendev.org/c/openstack/kayobe/+/93473112:03
PrzemekKMaybe You can teach about "From zero to code: A hands-on guide to making your first OpenStack code contribution" from kolla-ansible  side ? It would be interested how we can propose changes to master code and do local adjustments for example to images/ code https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/EYGHAMW765EFCXD7JLVYVBCHINPZP3ZS/ 12:30
opendevreviewMickael Razzouk proposed openstack/kolla master: Added libvirt dependency to nova-libvirt container  https://review.opendev.org/c/openstack/kolla/+/93495212:51
opendevreviewRafal Lewandowski proposed openstack/kolla-ansible master: Enable ML2/OVN and distributed FIP by default.  https://review.opendev.org/c/openstack/kolla-ansible/+/90495913:00
opendevreviewJakub Darmach proposed openstack/kolla-ansible master: Exclude NFS backend mount from ownership change  https://review.opendev.org/c/openstack/kolla-ansible/+/92394513:27
SvenKieskebbezak: meeting? :)14:00
bbezakcertainly ;)14:00
bbezak#startmeeting kolla14:00
opendevmeetMeeting started Wed Nov 13 14:00:34 2024 UTC and is due to finish in 60 minutes.  The chair is bbezak. Information about MeetBot at http://wiki.debian.org/MeetBot.14:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:00
opendevmeetThe meeting name has been set to 'kolla'14:00
bbezak#topic rollcall14:00
*** bbezak is now known as Guest930614:01
SvenKieskeo/14:01
frickler\o14:01
Guest9306I can see that core team is around, good :)14:02
Guest9306#topic agenda14:03
Guest9306* Roll-call... (full message at <https://matrix.org/oftc/media/v1/media/download/Aaudh7XK0DL4kQY_zkWsYCLPncW_1ApJNdaQnSNYQH0FUwwGMopjXbBNAYF1ZXJN3K2wNqjWndpaEAzyoxh2jx5CeTbe8zCAAG1hdHJpeC5vcmcvbmtpdEZJUURZdVlObk13eExmaVFsU1Jj>)14:03
Guest9306(i hope rendering is not very bad)14:03
fricklerit is14:03
SvenKieskerendering is very bad (just a link) :D14:03
frickleryou need to paste each line on it's own14:04
fricklerbut for now I guess we can live with it14:04
Guest9306yeah. I'll try to remember that14:04
Guest9306#topic CI status14:04
Guest9306whiteboard says green14:04
Guest9306I didn't see massive issues recently14:05
fricklerthere were some infra issues yesterday, but should be fine now again14:05
Guest9306ok14:05
Guest9306#topic Release tasks14:06
Guest9306how is kayobe doing14:06
Guest9306jovial14:06
Guest9306I'll check with him14:06
Guest9306#topic Current cycle planning14:07
Guest9306I guess we're just after PTG and in the beginning of the cycle14:08
Guest9306rabbitmq is a good topic, but I saw discussion in the change SvenKieske14:08
SvenKieskeyes, so on one hand CI is "green" but looking at actual logs it's not that green and I wonder why that many errors are reported as green, but I didn't have time to dive deep into the issues and if they are even related.14:09
Guest9306any particular example?14:10
SvenKieskeand if I'm not missing something it's really weird that CI is green, as afaik upstream removed some of our used queue types, so I wonder how that should work.14:10
SvenKieske#link https://zuul.opendev.org/t/openstack/build/0037175825154a0bb76e0c6c65eab14414:10
SvenKieskespecifically https://zuul.opendev.org/t/openstack/build/0037175825154a0bb76e0c6c65eab144/log/secondary1/logs/kolla/all-ERROR.txt but it's a very very long error log so as I said I didn't deep dive into it14:11
SvenKieskethere's errors all over the place, no sql connection, keystone domain not found..various stuff, might be related to rmq, might be something else14:12
Guest9306might be related to bootstrap. I'd compare other changes flows14:12
SvenKieskenot sure why that's reported as "success" though. afaik in k-a that should be a hard error. maybe zuul ci in kolla is different than in kolla-ansible? not sure.14:13
fricklerlot's of those errors are during the startup phase, so they are ignored14:13
Guest9306yes, we're ignoring a lots of errors. recently we added some as well14:13
fricklerand keystone domain/user/whatever not found is normal ops14:14
Guest9306in check-logs.sh14:14
Guest9306tests/check-logs.sh that is14:14
Guest9306sorry for autoparsing link14:14
SvenKieskeit's at least hard to distinguish multinode jobs from single node jobs in kolla, these seem to be differently labeled from the kolla-ansible jobs somehow14:14
SvenKieskebecause I don't think we would find any issues in single node jobs.14:15
Guest9306yes, multinodes are very helpful in that regard14:15
SvenKieskeor this: 2024-11-12 14:28:53.801 31 ERROR oslo_service.periodic_task oslo_messaging.exceptions.MessageDeliveryFailure: Unable to connect to AMQP server on 192.0.2.1:5672 after inf tries: Basic.publish: (404) NOT_FOUND - no exchange 'cinder-scheduler_fanout' in vhost '/'14:16
SvenKieskebasically last line in: https://zuul.opendev.org/t/openstack/build/19efb602a7734f06b495a0b2346bf345/log/secondary2/logs/kolla/all-ERROR.txt14:16
SvenKieskeso that doesn't look like a healthy rmq to me14:16
SvenKieskestill reported as success. I'm not sure what's filtering out "harmless" errors in our CI, but I think it's a little bit greedy14:17
SvenKieskeand in general I would be very careful about filtering out errors at all, because the required patterns often spin out of control and tend to shadow real issues.14:18
Guest9306indeed. however some errors in bootstrap will occur anyway14:19
SvenKieskebut whatever, someone would have to dedicate some time, maybe I can make some on sunday, to look into it, no promises though14:19
Guest9306ok. let's continue discussion in that change14:19
Guest9306it make sense14:19
SvenKieskesure, I understand the very good reasons to hide some errors, we can't fix all of them. we just need to be very careful :)14:19
Guest9306to make it clearer14:19
Guest9306#topic Additional agenda (from whiteboard)14:20
SvenKieskeI'll update the link to the latter link, that seems like a more real issue, thx14:20
Guest9306I can see some changes to review from SvenKieske 14:20
Guest9306and backports14:21
Guest9306that is on my list14:21
Guest9306I'll take a look into that14:21
SvenKieskenice, yeah most of those are backports I think :)14:22
SvenKieskeno reason to go over them individually imho. would just be happy to receive some reviews, thx.14:22
Guest9306yeap, will take a look14:23
Guest9306#topic Open discussion14:23
Guest9306I guess we did some open discussion already :)14:24
Guest9306I think we won't have meeting next week, as there are at least couple of conferences14:25
SvenKieskeokay14:25
Guest9306I'll be in Prague on OVS/OVN con14:25
Guest9306and some folks will be on Supercomputing in the US14:25
fricklerlet's skip it then, I'd think14:25
Guest9306I'm wondering why today's attendance is so low ;)14:25
SvenKieskecan you motivate someone at ovn con to revive the ovn-exporter project, please? ;)14:26
Guest9306:)14:26
Guest9306I'll try14:26
SvenKieskeupstream is dead and I don't feel particular inclined to maintain my one-off-patch fork, which isn't even finished :D14:26
Guest9306it is a problem of one-people project. I think ovs/ovn should have and maintain that kind of stuff14:27
SvenKieskebtw, your IRC name somehow changed to "Guest9306" just if you're wondering bbezak :D14:27
Guest9306lovely14:27
Guest9306again :)14:28
SvenKieskeright, would be nice if ovn/ovs project could step up. no idea how many ressources they got though.14:28
*** Guest9306 is now known as bbezak14:28
SvenKieskethat looks better, but I guess we are finished? :)14:29
bbezakI don't have anything else14:29
bbezaklet's wrap up then, thank you all!14:29
bbezak#endmeeting14:29
opendevmeetMeeting ended Wed Nov 13 14:29:57 2024 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)14:29
opendevmeetMinutes:        https://meetings.opendev.org/meetings/kolla/2024/kolla.2024-11-13-14.00.html14:29
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/kolla/2024/kolla.2024-11-13-14.00.txt14:29
opendevmeetLog:            https://meetings.opendev.org/meetings/kolla/2024/kolla.2024-11-13-14.00.log.html14:29
fricklero/14:30
SvenKieskeo/14:30
SvenKieskethx for running the meeting bbezak14:31
bbezak;)14:31
greatgatsbyHello.  I did another test upgrade from Zed to Antelope and we experienced another 5 minute outage to our VMs via the floating IPs.  This time we had kolla-ansible logging with timestamps and the outage occurred just as the "Restart openvswitch-db server container" and "Restart openvswitchd container" tasks were happening.  Really hoping there's something we can do to prevent this, appreciate any tips or help.  Thanks.15:24
opendevreviewAntony Messerli proposed openstack/kolla-ansible master: Enables override of Octavia notification topics  https://review.opendev.org/c/openstack/kolla-ansible/+/93482715:33
xiaolonghello15:44
xiaolongI encountered an issue regarding deploying OpenStack using Kolla15:44
xiaolongWhen deploying OpenStack 2024.1 and 2024.2 using CentOS 9 Steam and Rocky 9.415:45
xiaolongBootstrap servers' prompts error when installing 'dbus python'15:47
xiaolongIs 2024.2 really applicable to the Rocky9.4 system? The system's built-in Python 3.9 cannot even install Ansible core>=2.1615:49
SvenKieskethere was a fix regarding the dbus package but that should been included in 2024.1 and .216:02
SvenKieskehttps://review.opendev.org/c/openstack/ansible-collection-kolla/+/92261416:02
xiaolongDo you have a document address?16:02
xiaolongThank You16:03
SvenKieskenot so sure about your rocky issue, afaik kolla should take care of the appropriate python version inside the venv16:04
SvenKieskemaybe above dbus fix is not enough for all scenarios, but you should be able to adapt it to your scenario, bug reports are welcome, if you can confirm something is missing :)16:05
xiaolongok16:06
xiaolongIs there an article about deploying OpenStack starting from installing the system?16:07
SvenKieskemost assume you have a basic operating system installed somewhere I think16:07
SvenKieskeI think our downstream guide is probably one of the best which also mentions the operating system installation and prerequisites, but the disadvantage is, that it assumes you use our open source kolla wrapper: https://osism.tech/docs/guides/deploy-guide/seed16:08
SvenKieskexiaolong: I assume you have read https://docs.openstack.org/kolla-ansible/latest/user/quickstart.html already?16:10
xiaolongyes16:11
xiaolongBut there is no detailed explanation of the system version, such as whether it is written as Rocky9 in Quick Start, 9.0, 9.2, or 9.4?16:11
xiaolongrocky16:11
SvenKieskemhm all should work but from what I still know about rocky, afaik only their latest release is supported, no? so if 9.4 is released 9.3 is out of support? at least that was the case with centos some years ago.16:12
SvenKieskeyes, only 9.4 is supported: https://wiki.rockylinux.org/rocky/version/#__tabbed_1_216:12
SvenKieske9.3 is EOL since May 202416:13
xiaolongyes16:13
xiaolongOSISM may be useful in data center deployment, but I want to deploy OpenStack on a laptop, and OSISM requires RAM>=1616:13
SvenKieskeah okay16:13
SvenKieskewe have a special deployment method for laptops actually, but let me check the ram requirements16:13
xiaolongreally?16:14
SvenKieskeyeah, that even needs 32 GB of ram, due to ceph, unfortunately :(16:14
xiaolongdevstack?16:14
xiaolong......16:14
SvenKieskeso it's a complete cloud, including ceph: https://osism.tech/docs/guides/other-guides/cloud-in-a-box/16:14
SvenKieskeyou can make it work with 8GB if you don't start any payloads I think :D16:15
xiaolongYes, I think it must work very well in a production environment16:15
SvenKieskethere was a blog where someone made it run on his laptop16:15
SvenKieskeah yeah, but he had also 64 GB of ram: https://scs.community/2023/08/04/ciab-on-a-laptop/ but it's quite educational xiaolong. the next best thing would be devstack I think16:17
xiaolonghahaha16:19
xiaolongyes16:19
xiaolongHave you ever deployed OpenStack on VMware using the Quick Start feature?16:23
xiaolonghello17:02
xiaolongHave you ever deployed OpenStack on VMware using the Quick Start feature?17:02

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