| opendevreview | wangxiyuan proposed openstack/kolla-ansible master: [WIP]Add openEuler Distro support https://review.opendev.org/c/openstack/kolla-ansible/+/830115 | 01:16 |
|---|---|---|
| opendevreview | Dr. Jens Harbott proposed openstack/kolla-ansible stable/xena: WIP: Test setting a FQDN as hostname https://review.opendev.org/c/openstack/kolla-ansible/+/835002 | 06:33 |
| edebeste | Hi everyone, am I correct in assuming that hosts of different architecture are not supported for a multinoide kolla-ansible deploy? | 06:56 |
| edebeste | I'm trying to deploy a home stack using a rpi4 as a controller and an x86 machine as my hypervisor using Xena. In my inventory file I specify the container image suffix of the controller should be "-aarch64" (using the debian image). Most things seem to go well until I get to the nova-cells playbook that delegates from the hypervisor node to the | 07:02 |
| edebeste | controller where it appears that the incorrect image is being used to run the commands. I assume this is due to ansible using the host facts for the hypervisor node (that does not have the suffix for aarch64). I tried to use delegate_facts: true, but that did not seem to help | 07:02 |
| opendevreview | wangxiyuan proposed openstack/kolla-ansible master: [WIP]Add openEuler Distro support https://review.opendev.org/c/openstack/kolla-ansible/+/830115 | 07:49 |
| opendevreview | Dr. Jens Harbott proposed openstack/kolla-ansible master: designate: allow designate_ns_record to be a list https://review.opendev.org/c/openstack/kolla-ansible/+/802304 | 07:51 |
| yoctozepto | edebeste: hi! it was never tested so can be assumed as "not supported" but I think it is something that we might want to support; thus, you may want to report a bug with details on which task fails and how (plus the inventory and config details) | 07:54 |
| edebeste | yoctozepto, thanks for the confirmation. I'll submit a report later and may even try to dig into the ansible as well... have to read the how to contribute guide first :P | 07:55 |
| yoctozepto | :D | 07:55 |
| frickler | yoctozepto: mnasiadka: can we still get 802304 into yoga? in the current state, it is backwards compatible, so might even work without a reno? | 08:07 |
| yoctozepto | frickler: will check later | 08:08 |
| mnasiadka | speaking about designate - this one would be nice as well - https://review.opendev.org/c/openstack/kolla-ansible/+/802301 | 08:10 |
| frickler | regarding the libvirt sasl issue, I've added debugging that shows that in the k-a CI the right thing is happening, the user entry is created correctly as user@fqdn. now I'll have to test again downstream and find out why it is not working there | 08:15 |
| yoctozepto | frickler: /etc/{hostname,hosts} unaligned? | 08:32 |
| frickler | yoctozepto: good point, /etc/hostname indeed seems to contain the short hostname | 08:58 |
| opendevreview | wangxiyuan proposed openstack/kolla-ansible master: [WIP]Add openEuler Distro support https://review.opendev.org/c/openstack/kolla-ansible/+/830115 | 09:02 |
| hu_berlin_kalle | Hi, I'm trying to deploy kayobe and multiple nodes. I've successfully run kayobe overcloud service deploy. But my floating ip net has no connectivity. When I look into my ml2.conf I notice there is no directive connecting physnet to some real interface. | 09:16 |
| hu_berlin_kalle | Question: is that normal and can I deploy anything else but openvswitch via Kayobe? | 09:17 |
| hu_berlin_kalle | (like linuxbridge or ovn) | 09:17 |
| hu_berlin_kalle | kayobe wallaby/stable I might add | 09:18 |
| mnasiadka | ovn yes, linuxbridge - maybe (haven't tried) | 09:31 |
| hu_berlin_kalle | how do i configure that? | 09:41 |
| hu_berlin_kalle | do i just out it in kayobe_configpath/kolla/globals.yml? | 09:41 |
| hu_berlin_kalle | *put | 09:42 |
| hu_berlin_kalle | and do i then need to provide an ml2.conf.ini? | 09:42 |
| hu_berlin_kalle | or should kayobe still do that | 09:42 |
| hu_berlin_kalle | (from what it looks to me the template for ml2.conf is missing an openvswitch section) | 09:43 |
| hu_berlin_kalle | which is why i'm looking for an alternative | 09:44 |
| mnasiadka | https://docs.openstack.org/kolla-ansible/latest/reference/networking/neutron.html#ovn-ml2-ovn for Kolla-Ansible (so just put those in kolla/globals.yml) or kolla_enable_ovn in kolla.yml (that will set neutron_plugin_agent: "ovn") but not the additional vars | 09:46 |
| mgoddard | hu_berlin_kalle: openvswitch should work | 09:46 |
| mgoddard | hu_berlin_kalle: ovn should also work | 09:46 |
| mgoddard | hu_berlin_kalle: we don't generally test linuxbridge in kayobe, but it might work | 09:46 |
| hu_berlin_kalle | meaning there should be an ovs section in my ml2.conf just as i expected richgt? | 09:47 |
| hu_berlin_kalle | *right | 09:47 |
| mnasiadka | IIRC CentOS 8 had some weirdness blocking linuxbridge working, but don't remember the details | 09:47 |
| mgoddard | hu_berlin_kalle: https://opendev.org/openstack/kolla-ansible/src/branch/master/ansible/roles/neutron/templates/ml2_conf.ini.j2 | 09:48 |
| hu_berlin_kalle | thank I'm trying to deploy with linuxbridge on centos 8 stream right now | 09:48 |
| mgoddard | hu_berlin_kalle: https://opendev.org/openstack/kolla-ansible/src/branch/master/ansible/roles/neutron/templates/openvswitch_agent.ini.j2 | 09:48 |
| mgoddard | that is where the ovs section ends up | 09:48 |
| mgoddard | hu_berlin_kalle: TBH I wouldn't recommend linuxbridge, as we don't test it | 09:49 |
| hu_berlin_kalle | I'm really just desperate trying that :) | 09:49 |
| mnasiadka | ML2/OVS should work out of the box, if it doesn't - something went really wrong | 09:50 |
| hu_berlin_kalle | I'de rather go with ovn or ovs (just don't have much experience with it) | 09:50 |
| hu_berlin_kalle | so I'll toss the linuxbridge deployment right away. | 09:51 |
| hu_berlin_kalle | But I also didn't see the kolla_enable_onv switch so I'll try that one. | 09:52 |
| hu_berlin_kalle | when I put neutron_plugin_agent: "ovn" into kayobe_config_path/kolla/globals.yml I ended up with neutron_plugin_agent: "openvswitch" in the generated /etc/kolla/glogals.yml but also an ovn section in the generated ml2.conf | 09:55 |
| hu_berlin_kalle | And the I wasn't sure what happened ... | 09:55 |
| yoctozepto | mgoddard: hi! any progress on ironic patches reviews? :-) | 09:56 |
| mgoddard | yoctozepto: not yet. In the queue | 09:57 |
| yoctozepto | ack | 09:57 |
| hu_berlin_kalle | and one side question: To me it looks like any network interface that I want to assign to any network role has to be a bridge interface. BEcause otherwise I'll loose connectivity once ovs is plugged on top of it. Is that correct? | 10:01 |
| yoctozepto | frickler: you can do your first w+1 in kolla https://review.opendev.org/c/openstack/kolla-ansible/+/834648 | 10:13 |
| opendevreview | Michal Nasiadka proposed openstack/kolla-ansible master: designate: Allow to disable notifications https://review.opendev.org/c/openstack/kolla-ansible/+/802301 | 10:17 |
| mnasiadka | yoctozepto: you also have a feeling nobody is reading renos? :) | 10:19 |
| yoctozepto | mnasiadka: yup, pretty much, plus it takes effort to get renos right - commit messages should be good enough, I have added this topic to the PTG agenda | 10:20 |
| yoctozepto | I consider our action a "nice try but let's drop it" thing | 10:20 |
| mnasiadka | I was thinking about not doing version tagged releases, just stable branches (like stable/yoga), but have no clue how many people just do pip install kolla-ansible ;-) | 10:21 |
| mgoddard | hu_berlin_kalle: if you want to use the interface for anything else, it should be a bridge | 10:23 |
| mgoddard | mnasiadka, yoctozepto: I vote to keep renos, but be less perfectionist about them | 10:24 |
| mnasiadka | that also is an option | 10:24 |
| mgoddard | let's spend less time nit picking on grammar and formatting, just ensure the right info is present | 10:25 |
| yoctozepto | mnasiadka, mgoddard: i.e., ignore if missing? | 10:25 |
| mnasiadka | for bugs essentially we could just list bug number that is being closed in the reno, for the rest - require it only if there's really impact for existing deployment | 10:25 |
| yoctozepto | indeed | 10:25 |
| mgoddard | yoctozepto: we might discuss the threshold for requiring a reno, but in general still require one | 10:25 |
| yoctozepto | the threshold is tricky | 10:26 |
| yoctozepto | but right, for upgrade renos there really is no better way to convey that info | 10:26 |
| yoctozepto | hard topic | 10:26 |
| opendevreview | Sven Kieske proposed openstack/kolla-ansible master: fix missing lang env in curator crontab https://review.opendev.org/c/openstack/kolla-ansible/+/781114 | 10:27 |
| mgoddard | If I thought it would work, I'd suggest redirecting the time towards docs... | 10:27 |
| yoctozepto | yeah, the docs need love | 10:28 |
| yoctozepto | always | 10:28 |
| mgoddard | If you consider renos as ephemeral, vs code and docs as long-lived, it shows where to put the time | 10:29 |
| mgoddard | (in terms of perfectionism) | 10:29 |
| opendevreview | Sven Kieske proposed openstack/kolla-ansible master: fix missing lang env in curator crontab https://review.opendev.org/c/openstack/kolla-ansible/+/781114 | 10:29 |
| hu_berlin_kalle | mgoddard: thanks! Not sure weather that's super obvious. I would recommend some section in the documentation about that in both the config guide and the deployment guide. (Happy to provide it myself if that would help.) | 10:29 |
| mnasiadka | maybe we could talk about it on the PTG, like define what would be ideal to have in the per-project/scenario docs - I was thinking about listing related variables user can set, instead of leaving most of the users to code crawling :) | 10:30 |
| yoctozepto | well, the renos are medium-long-lived, let's say; they are docs in their own | 10:30 |
| mgoddard | mnasiadka: we have talked docs to death. I'm unwilling to talk more without promise of action :D | 10:31 |
| mnasiadka | mgoddard: basically, if we work out something that is detailed how docs should look like, I can commit myself update all the reference docs (Neutron, etc) with that - but don't look at me at doing the non-reference ones (like quickstart and so on) :) | 10:33 |
| opendevreview | Merged openstack/kayobe stable/xena: Ubuntu: support host package update https://review.opendev.org/c/openstack/kayobe/+/834607 | 10:36 |
| mgoddard | hu_berlin_kalle: it's not super obvious. There is a hint in https://docs.openstack.org/kayobe/latest/configuration/reference/network.html#neutron-networking, but if you can improve on it that would be nice | 10:37 |
| yoctozepto | mnasiadka: reference docs are usually good enough, except where they are missing; the worse is the general architecture section imho | 10:40 |
| mnasiadka | yoctozepto: you mean 'production architecture guide'? | 10:41 |
| yoctozepto | mnasiadka: kind of as it's simply not anywhere :D | 10:42 |
| mnasiadka | basically yes... | 10:42 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: deploy libvirt on the host https://review.opendev.org/c/openstack/kayobe/+/825359 | 10:46 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: support SASL authentication https://review.opendev.org/c/openstack/kayobe/+/833023 | 10:46 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: Enable host daemon by default https://review.opendev.org/c/openstack/kayobe/+/835034 | 10:46 |
| mnasiadka | mgoddard: merge conflicts on the kolla collection patches, can you update? :) | 10:49 |
| mnasiadka | https://review.opendev.org/c/openstack/ansible-collection-kolla/+/821016 | 10:49 |
| hu_berlin_kalle | mgoddard: you have no idea how often I read that paragraph :D - I'll figure out how to contribute to the documentation and then make some suggestions. And thanks for your help! | 10:51 |
| mgoddard | hu_berlin_kalle: well, glad to know someone is reading it, even if it isn't that helpful :) | 10:52 |
| yoctozepto | :D | 10:54 |
| mnasiadka | well it's either 3 days of troubleshooting or 5 minutes of reading, even if it's not that helpful ;-) | 10:55 |
| sorin-mihai | what would be the recommended combination for a production HCI deployment with external ceph? what host OS, what ceph deployment tool, and which kolla containers? | 10:57 |
| opendevreview | Mark Goddard proposed openstack/ansible-collection-kolla master: baremetal: refactor docker_sdk installation to a separate role https://review.opendev.org/c/openstack/ansible-collection-kolla/+/821016 | 11:01 |
| opendevreview | Mark Goddard proposed openstack/ansible-collection-kolla master: baremetal: refactor docker deployment into a separate role https://review.opendev.org/c/openstack/ansible-collection-kolla/+/821204 | 11:01 |
| opendevreview | Mark Goddard proposed openstack/ansible-collection-kolla master: docker: restart docker and containerd in handlers https://review.opendev.org/c/openstack/ansible-collection-kolla/+/821205 | 11:01 |
| opendevreview | Mark Goddard proposed openstack/ansible-collection-kolla master: docker: add registry CA configuration https://review.opendev.org/c/openstack/ansible-collection-kolla/+/821206 | 11:01 |
| opendevreview | Mark Goddard proposed openstack/ansible-collection-kolla master: baremetal: refactor package installation into a separate role https://review.opendev.org/c/openstack/ansible-collection-kolla/+/829586 | 11:01 |
| opendevreview | Mark Goddard proposed openstack/ansible-collection-kolla master: baremetal: refactor libvirt apparmor configuration https://review.opendev.org/c/openstack/ansible-collection-kolla/+/829587 | 11:01 |
| mnasiadka | yoctozepto: another list to review ^^ | 11:10 |
| yoctozepto | oh my | 11:14 |
| mgoddard | sorin-mihai: whichever supported distro you are most comfortable with | 11:17 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: Enable host daemon by default https://review.opendev.org/c/openstack/kayobe/+/835034 | 11:24 |
| opendevreview | Merged openstack/kayobe stable/xena: Ubuntu: add support for Apt repository configuration https://review.opendev.org/c/openstack/kayobe/+/834847 | 11:34 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: deploy libvirt on the host https://review.opendev.org/c/openstack/kayobe/+/825359 | 11:40 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: support SASL authentication https://review.opendev.org/c/openstack/kayobe/+/833023 | 11:40 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: Enable host daemon by default https://review.opendev.org/c/openstack/kayobe/+/835034 | 11:40 |
| hu_berlin_kalle | in stable/wallaby in the template for kolla globals.yml neutron_plugin_agent is fixed to openvswitch. Source: https://opendev.org/openstack/kayobe/src/branch/stable/wallaby/ansible/roles/kolla-ansible/templates/globals.yml.j2 | 11:44 |
| hu_berlin_kalle | is this a bug? | 11:44 |
| opendevreview | Verification of a change to openstack/kayobe stable/xena failed: Ubuntu: add support for Apt configuration https://review.opendev.org/c/openstack/kayobe/+/834929 | 11:56 |
| opendevreview | wangxiyuan proposed openstack/kolla-ansible master: Add openEuler Distro support https://review.opendev.org/c/openstack/kolla-ansible/+/830115 | 12:01 |
| opendevreview | Merged openstack/kolla-ansible master: designate: fix external backend deployment https://review.opendev.org/c/openstack/kolla-ansible/+/834648 | 12:02 |
| opendevreview | Dr. Jens Harbott proposed openstack/kolla-ansible master: CI: IPv6 external network https://review.opendev.org/c/openstack/kolla-ansible/+/712768 | 12:04 |
| frickler | yoctozepto: ^^ very curious how that will work out | 12:04 |
| yoctozepto | me too | 12:09 |
| opendevreview | Radosław Piliszek proposed openstack/kolla-ansible stable/xena: designate: fix external backend deployment https://review.opendev.org/c/openstack/kolla-ansible/+/835050 | 12:10 |
| opendevreview | Radosław Piliszek proposed openstack/kolla-ansible stable/wallaby: designate: fix external backend deployment https://review.opendev.org/c/openstack/kolla-ansible/+/835051 | 12:10 |
| opendevreview | Radosław Piliszek proposed openstack/kolla-ansible stable/victoria: designate: fix external backend deployment https://review.opendev.org/c/openstack/kolla-ansible/+/835052 | 12:11 |
| SvenKieske | mnasiadka|frickler: could you please sort out, which language you want to have here? you got conflicting requirements :) https://review.opendev.org/c/openstack/kolla-ansible/+/781114 | 12:46 |
| opendevreview | Merged openstack/kolla-ansible master: designate: allow designate_ns_record to be a list https://review.opendev.org/c/openstack/kolla-ansible/+/802304 | 12:52 |
| frickler | SvenKieske: well maybe my wording was bad, but I didn't explictly require C.UTF-8, I was just wondering what was the reason to prefer en_US.UTF-8. that it is already used elsewhere is a good enough reason for me, so I'm fine with keeping the variant. I can push the old PS for you if you want | 13:02 |
| yoctozepto | SvenKieske: which release you know is broken? | 13:11 |
| yoctozepto | and which distro | 13:11 |
| watz | hello everybody | 13:18 |
| SvenKieske | yoctozepto: I know of train/ubuntu (it should be mentioned in the linked bug report imho). I don't know enough about crontab and elasticsearch curator cli to assess if other distros are impacted, but the steps to reproduce the problem are as well mentioned in the bug report | 13:22 |
| SvenKieske | as this was never patched this also applies to master | 13:23 |
| yoctozepto | interesting, it works for me on wallaby/debian | 13:23 |
| SvenKieske | I honestly have no energy atm to test on other plattforms, but the reproducer is really straight forward, just docker exec; unset LANG, execute the script, watch python trace (or not) | 13:23 |
| SvenKieske | maybe debian install scripts do some magic with regard to env stuff? :shrugs: (would not be the first time) | 13:24 |
| yoctozepto | nah, they have newer python which I believe always defaults to unicode | 13:25 |
| SvenKieske | it _is_ weird, that is true :) | 13:25 |
| SvenKieske | ah, interesting! :) | 13:25 |
| watz | is there an easy way to migrate databases to a newer version? | 13:25 |
| SvenKieske | frickler: seems reasonable; I did forget myself, that en_US already was set via docker env, so yeah, seems to be the best to align with that. | 13:25 |
| SvenKieske | watz: can you be more specific? which database? mariadb openstack db? I believe that is handled during kolla upgrade? | 13:26 |
| watz | i have an old pike installation .. my testinstallation is now xena .. but i want to key image/volume information, which is stored in the mariadb, right? | 13:27 |
| opendevreview | Sven Kieske proposed openstack/kolla-ansible master: fix missing lang env in curator crontab https://review.opendev.org/c/openstack/kolla-ansible/+/781114 | 13:27 |
| watz | s/key/keep | 13:28 |
| SvenKieske | watz: yeah, sounds reasonable, but I would advise to do upgrades release by release and not to skip releases (in theory you should be able to skip, I believe, 2 releases). the db migrationscript run automatically when using kolla-ansible | 13:33 |
| SvenKieske | watz: and use kolla-ansible to perform a backup before upgrading, in case something goes wrong | 13:34 |
| watz | SvenKieske: jeah, i also came up with that idea :-) but upgrading my whole cluster would take a while then :-) | 13:35 |
| SvenKieske | watz: clarifying question: do you want to run an in place upgrade, or do you want to export the database into a new installation on different hosts? | 13:35 |
| watz | ok .. my "old" production cluster is ubuntu 16.04 with openstack pike .. the idea is to reinstall the whole cluster and upgrade ubuntu to 20.04 | 13:36 |
| SvenKieske | watz: I believe to go straight from pike to xena would lead to some breakage; but I'm all speaking theoretically here, because I never did an upgrade myself, we are in fact preparing our own first upgrades right now | 13:36 |
| watz | i have a small (4 nodes) testcluster running ubuntu 20.04 and a fresh install of xena with the help of kolla :-) | 13:37 |
| SvenKieske | watz: so maybe wait for someone with more production knowledge than me | 13:37 |
| watz | hehehehe :-) | 13:37 |
| watz | nice :-) | 13:37 |
| watz | since i use a ceph cluster and ldap auth as backend .. users and images/voluems are not the problem. i want to keep the relations between users/projects/images/volumes | 13:39 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: deploy libvirt on the host https://review.opendev.org/c/openstack/kayobe/+/825359 | 13:46 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: support SASL authentication https://review.opendev.org/c/openstack/kayobe/+/833023 | 13:46 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: Enable host daemon by default https://review.opendev.org/c/openstack/kayobe/+/835034 | 13:47 |
| SvenKieske | watz: that said, you might have luck on the users mailinglist, as IRC ist most of the time busy with dev related stuff, but ymmv | 13:51 |
| watz | SvenKieske: thx .. will try that :-) | 13:51 |
| mnasiadka | SvenKieske, yoctozepto: you want to say, that Sven's patch might be not fixing anything? ;-) | 14:28 |
| yoctozepto | mnasiadka: just not fixing debian which works | 14:29 |
| yoctozepto | maybe it really is train/ubuntu only that is broken | 14:29 |
| mnasiadka | parallax: have we ever observed curator not working/running/failing? | 14:31 |
| SvenKieske | mhm, so should I rework the patch to be only included in ubuntu builds? but would that be worth the hassle? | 14:34 |
| yoctozepto | SvenKieske: or maybe even only ubuntu before focal | 14:36 |
| yoctozepto | i.e., before victoria | 14:37 |
| yoctozepto | in regular (non-EM) support we have victoria+ | 14:38 |
| SvenKieske | but I understand that is speculation? or did you test on victoria/focal that it works? or are there tests for the curator which actually check, that it is running? last time I looked I didn't find any, but that might be due to my own inability. | 14:40 |
| yoctozepto | SvenKieske: speculation based on the fact that we have not noticed and we use it in prod | 14:42 |
| SvenKieske | well it _is_ definitely broken in train/ubuntu and I bet someone uses that in production and didn't notice either. we just noticed because a test env did run out of space and I checked why curator didn't work. because superficially it always "seems" to run, checking "docker ps" is not enough.. | 14:43 |
| SvenKieske | will double check, afaik base packages between ubuntu/debian should be the same, but would appreciate if someone could check some prod/dev env on newer ubuntu and/or openstack releases as I currently got none at hand. | 14:45 |
| yoctozepto | yeah, I mean it's properly cleaning up es where we look at | 14:45 |
| SvenKieske | so you see es storage usage going down after curator run? sorry to be so pedantic, but colleagues also assured me curator would run (and I thought so myself) until I ran the command manually via CLI the way the crontab user does which spew stacktraces at me | 14:46 |
| yoctozepto | no worries, I understand your digging all too well :D yes, it does | 14:48 |
| SvenKieske | mhm, weird. want to figure that one out, and will try to doublecheck that it's not our environment which introduced some weird bug (we build ubuntu source based containers and images with kolla and disk-image-builder) | 14:49 |
| yoctozepto | I believe it's not your env per se, just the used combination (train+ubuntu) | 14:49 |
| yoctozepto | I would bet ussuri+ubuntu is also b0rken but not later | 14:50 |
| yoctozepto | and also not concurrent distros | 14:50 |
| parallax | mnasiadka: yes, due to python libraries, that should be fixed now | 14:50 |
| yoctozepto | mnasiadka, parallax: but that was a different issue | 14:50 |
| yoctozepto | thanks parallax | 14:50 |
| parallax | sorry, I'm out of context :) | 14:51 |
| yoctozepto | indeed you are :-) | 14:51 |
| SvenKieske | well ussuri/ubuntu I can soon tell you, because we will upgrade soonish, then victoria as well | 14:53 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: libvirt: Enable host daemon by default https://review.opendev.org/c/openstack/kayobe/+/835034 | 14:54 |
| frickler | yoctozepto: doesn't look too bad I'd say, I'll fix the fip_addr calculation https://cfa62c8928af95e5821c-1a0f674a3837e35101e72ddbc77fd8ba.ssl.cf1.rackcdn.com/712768/9/check/kolla-ansible-ubuntu-source-multinode-ipv6/de50d81/primary/logs/ansible/test-core-openstack | 14:55 |
| yoctozepto | frickler: well, it says it failed to start dropbear (ssh daemon) | 14:57 |
| yoctozepto | and also that it has issues pinging the gateway | 14:58 |
| yoctozepto | so I'm not entirely sure it is not that bad ;-) | 14:58 |
| yoctozepto | as for me, play with it to your heart contents and then we will review | 14:58 |
| yoctozepto | it makes sense to test full ipv6 stack | 14:58 |
| yoctozepto | hmm, it seems the dropbear runs nonetheless, it's just a wrong message | 15:08 |
| yoctozepto | and the address is there so might be it works despite the gateway ping failure; the CI and cirros systems are directly connected via bridge so there is no gateway in play | 15:09 |
| sorin-mihai | in the ceph mon nodes there is a /etc/ceph/ceph.conf tha contains [client.libvirt], [client.$hostname.rgw0] and [global] sections; all node seem to have the same sections, slight differences. are all those sections needed in /etc/kolla/config for the services that need to connect to ceph or the [global] section is enough? | 15:21 |
| * yoctozepto off | 15:26 | |
| opendevreview | Dr. Jens Harbott proposed openstack/kolla-ansible master: CI: IPv6 external network https://review.opendev.org/c/openstack/kolla-ansible/+/712768 | 15:29 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: Pin Jinja2<3.1.0 to avoid Python 3.6 drop https://review.opendev.org/c/openstack/kayobe/+/835089 | 15:57 |
| jamesbenson | I was pulling images on xena and got this error this morning: `FAILED! => {"msg": "template error while templating string: No filter named 'select_services_enabled_and_mapped_to_host'.. String: {{ lookup('vars', (kolla_role_name | default(project_name)) + '_services') | select_services_enabled_and_mapped_to_host }}"}` | 15:59 |
| opendevreview | Mark Goddard proposed openstack/kolla-ansible master: Pin Jinja2<3.1.0 to avoid Python 3.6 drop https://review.opendev.org/c/openstack/kolla-ansible/+/835090 | 16:00 |
| opendevreview | Rafael Weingartner proposed openstack/kolla-ansible master: Customize the authentication error timeout page in modOIDC https://review.opendev.org/c/openstack/kolla-ansible/+/832806 | 16:03 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: Pin Jinja2<3.1.0 to avoid Python 3.6 drop https://review.opendev.org/c/openstack/kayobe/+/835089 | 16:06 |
| frickler | jamesbenson: seems to be the issue that mgoddard just posted workarounds for | 16:14 |
| frickler | though I'd prefer if we could fix this properly and pin only for historic distros | 16:15 |
| mgoddard | frickler: 1. put fire out | 16:18 |
| mgoddard | frickler: 2. move forward | 16:18 |
| opendevreview | Verification of a change to openstack/kayobe stable/xena failed: Ubuntu: add support for Apt configuration https://review.opendev.org/c/openstack/kayobe/+/834929 | 16:24 |
| opendevreview | Mark Goddard proposed openstack/kolla-ansible master: Use jinja2.pass_context instead of contextfilter https://review.opendev.org/c/openstack/kolla-ansible/+/835092 | 16:26 |
| opendevreview | Mark Goddard proposed openstack/kolla-ansible master: Use jinja2.pass_context instead of contextfilter https://review.opendev.org/c/openstack/kolla-ansible/+/835092 | 16:27 |
| mgoddard | frickler: I think you're right, the fix is as simple as the workaround | 16:28 |
| mgoddard | frickler: no pin required it seems, they did the setup.py/cfg properly | 16:28 |
| opendevreview | Mark Goddard proposed openstack/kolla-ansible master: Use jinja2.pass_context instead of contextfilter https://review.opendev.org/c/openstack/kolla-ansible/+/835092 | 16:30 |
| opendevreview | Merged openstack/kolla-ansible master: Enable memcached backend for mod_auth_openidc https://review.opendev.org/c/openstack/kolla-ansible/+/794609 | 16:30 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: Use jinja2.pass_context instead of contextfilter https://review.opendev.org/c/openstack/kayobe/+/835093 | 16:34 |
| frickler | mgoddard: you still may need the pin for py3.6 in CI because of our wheel mirror, it doesn't serve the necessary metadata | 16:34 |
| mgoddard | hmm | 16:35 |
| mgoddard | let's see | 16:35 |
| mgoddard | should fail quickly | 16:35 |
| * frickler afks for a bit, will check results later | 16:38 | |
| opendevreview | Sven Kieske proposed openstack/kolla-ansible master: re-add rabbitmq config for clustering interface https://review.opendev.org/c/openstack/kolla-ansible/+/758576 | 16:53 |
| opendevreview | Juan Pablo Suazo proposed openstack/kolla master: Adds the vitrage-snmp-parsing service. https://review.opendev.org/c/openstack/kolla/+/833628 | 16:56 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: Use jinja2.pass_context instead of contextfilter https://review.opendev.org/c/openstack/kayobe/+/835093 | 17:09 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: Pin Jinja2<3.1.0 to avoid Python 3.6 drop https://review.opendev.org/c/openstack/kayobe/+/835089 | 17:10 |
| opendevreview | Mark Goddard proposed openstack/kayobe master: Pin Jinja2<3.1.0 to avoid contextfilter removal https://review.opendev.org/c/openstack/kayobe/+/835089 | 17:11 |
| opendevreview | Rafael Weingartner proposed openstack/kolla-ansible master: Customize the authentication error timeout page in modOIDC https://review.opendev.org/c/openstack/kolla-ansible/+/832806 | 18:17 |
| opendevreview | Merged openstack/kolla-ansible master: Use jinja2.pass_context instead of contextfilter https://review.opendev.org/c/openstack/kolla-ansible/+/835092 | 21:31 |
| opendevreview | Maksim Malchuk proposed openstack/kolla-ansible stable/xena: Use jinja2.pass_context instead of contextfilter https://review.opendev.org/c/openstack/kolla-ansible/+/835126 | 21:34 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!