| opendevreview | Andrew Bonney proposed openstack/openstack-ansible master: Remove dynamic-address-fact call which is no longer required https://review.opendev.org/c/openstack/openstack-ansible/+/890306 | 08:35 |
|---|---|---|
| mgariepy | anyone here is using stslite with keystone ? : https://docs.ceph.com/en/quincy/radosgw/STSLite/ ? | 13:04 |
| lsudre | Hello, OSA installation succeeded. As admin, error message "Something went wrong!" like this bug: https://bugs.launchpad.net/openstack-ansible/+bug/2002897. Same issue in syslog "No type driver for tenant network_type: vxlan" when modifying ml2_conf.ini with "vlan,geneve,flat", syslog error: "Unrecoverable error: please check log for details.: ValueError: :6642: bad peer name format". How to fix this bug? Thank you | 13:35 |
| mgariepy | lsudre, which branch/tag ? | 13:38 |
| mgariepy | this issue has been fixed since a long time... | 13:38 |
| jrosser | mgariepy: i think we did some stuff with stslite | 13:42 |
| lsudre | mgariepy: 27.0.1 | 13:42 |
| spatel | I am getting this error when running - openstack coe cluster config mycluster-31 | 13:42 |
| spatel | a bytes-like object is required, not 'str' | 13:42 |
| spatel | does anyone know what is wrong? | 13:42 |
| spatel | https://paste.opendev.org/show/btg71oY0lTcE8UwM74dq/ | 13:43 |
| mgariepy | is it only setting the 2 config keys or needs to adjust policy or something ? | 13:43 |
| mgariepy | spatel, i don't use magnum | 13:43 |
| spatel | lucky you :) | 13:43 |
| mgariepy | spatel, it's not luck ;p | 13:44 |
| jrosser | we don;t use it either | 13:44 |
| jrosser | becasue we tried, and it was unmaintainable | 13:44 |
| opendevreview | Andrew Bonney proposed openstack/openstack-ansible-os_ironic master: Fix lack of constraints for ironic-inspector https://review.opendev.org/c/openstack/openstack-ansible-os_ironic/+/890335 | 13:44 |
| spatel | its mess to run k8s | 13:45 |
| mgariepy | unmaintainability is somewhat an issue with a log of stuff ;) quite often stuff aroung k8s... lol | 13:45 |
| spatel | my company pushing to use k8s on openstack to reduce cost from GCP :D | 13:46 |
| jrosser | i had someone on my team spent pretty much the whole rocky release cycle getting magnum to a point where it worked in our lab | 13:46 |
| jrosser | everything changed in stein and that whole work was trashed | 13:46 |
| spatel | I am running Xena but still its mess | 13:46 |
| spatel | so many version and only specific version works | 13:47 |
| jrosser | well you probably want to talk to mnaser ^^ | 13:47 |
| mnaser | spatel: we built a whole new solution that solves this forever :) | 13:47 |
| jrosser | ^ there you go :) | 13:47 |
| mnaser | https://github.com/vexxhost/magnum-cluster-api cluster api driver for magnum | 13:47 |
| mgariepy | wow | 13:47 |
| spatel | haha!! mnaser is k8s lover. In last summit he was all over on k8s :D | 13:48 |
| mnaser | instead of relying on the mess that is heat/magnum/fedora coreos, you would simply build images using image-builder | 13:48 |
| jrosser | i have this nearly working in an OSA AIO right now | 13:48 |
| mnaser | (kubernetes-sig project) | 13:48 |
| mnaser | and then just upload them to glance and let cluster api (built and developed by the fine k8s-sig folks) take care of it | 13:48 |
| lsudre | mgariepy: github 27.0.1 tag is newer than the bug old, why I have this bug? | 13:48 |
| mnaser | bootstrapped by kubeadm, running on ubuntu, life is good | 13:49 |
| mgariepy | lsudre, maybe the config is not quite ok ? | 13:49 |
| mgariepy | lsudre, can you post it somewhere? | 13:49 |
| spatel | mnaser I think documentation is little challenge. | 13:49 |
| mnaser | spatel: https://vexxhost.github.io/magnum-cluster-api/user/getting-started/ | 13:50 |
| mnaser | https://vexxhost.github.io/magnum-cluster-api/admin/intro/ | 13:50 |
| mnaser | and if you want to warm up to it and see how it works - https://vexxhost.github.io/magnum-cluster-api/developer/testing-and-development/ | 13:50 |
| mnaser | jrosser almost has this working inside of osa as well, so we'll hopefully have docs and playbooks for that | 13:50 |
| lsudre | mgariepy: ofc | 13:51 |
| spatel | Hmm! let me give it a shot.. | 13:51 |
| lsudre | mgariepy: here: https://paste.openstack.org/show/bxsX2leuA72UJheV3x79/ | 13:52 |
| mnaser | spatel: try it with the hack.sh script first which will deploy with devstack, you'll need a bm system or a vm with nested virt -- i think you'll like the simplicity of the experience | 13:52 |
| spatel | mnaser ok sounds good to me. I do have BM in lab so let me try and see how it goes | 13:53 |
| mnaser | spatel: ok cool, we're getting a slack channel setup on the k8s slack soon -- but feel free to ping | 13:54 |
| mgariepy | lsudre, also the user config please. | 13:54 |
| spatel | slack would be great! Please let me know when you have that channel ready. Sure I will ping you when I need you. | 13:54 |
| spatel | Thanks! mnaser | 13:54 |
| lsudre | mgariepy: user_variable.yml? | 13:55 |
| mgariepy | yep | 13:55 |
| lsudre | mgariepy: here: https://paste.openstack.org/show/b8CEreZLjzFIYacp87XN/ | 13:56 |
| lsudre | mgariepy: l20 to 23 are commented, this is the bug? | 13:56 |
| mnaser | spatel: you can keep an eye on https://github.com/kubernetes/community/pull/7418 -- but we are also in the #atmosphere slack channel on k8s slack if you're curious too | 14:00 |
| spatel | Thank for the heads up! I will join that slack channel | 14:02 |
| mgariepy | you probably need to change the opesntack_user_config.yml to reflect the ovn stuff. | 14:09 |
| mgariepy | lsudre, https://docs.openstack.org/openstack-ansible-os_neutron/latest/app-ovn.html | 14:10 |
| mgariepy | documentation need some love .. | 14:10 |
| lsudre | ok and after that, should I re-run every setup-xxx.yml scripts? | 14:10 |
| mgariepy | probably the neutron part will need to be run . | 14:16 |
| tuggernuts1 | when I run setup-openstack I get this error https://pastebin.com/HATBVvEA but when I run it again it seems to move on | 14:44 |
| tuggernuts1 | is there a fix for this? | 14:44 |
| jrosser | tuggernuts1: it would be good to know which service deployment causes this | 14:46 |
| jrosser | it's impossible to say where that as failed with only one task in the paste, can you provide some more output | 14:47 |
| tuggernuts1 | this looks like it's keystone | 14:47 |
| jrosser | ok well thats useful, becasue keystone is the first service to deploy | 14:48 |
| tuggernuts1 | did you want the full output? | 14:48 |
| jrosser | so that indicates that there is some connectivity issue with your db and haproxy | 14:48 |
| tuggernuts1 | ah ok | 14:48 |
| jrosser | first, take a look at the haproxy status and see if the backend for galera db is up | 14:49 |
| tuggernuts1 | might explain why it comes up the second time | 14:49 |
| tuggernuts1 | could it be the ha doesn't see the db is up before it run this part of the deploy? | 14:49 |
| jrosser | it should detect that quickly | 14:50 |
| tuggernuts1 | kk, let me dig on this one thanks | 14:50 |
| jrosser | also you can see here that logging is suppressed for those db_setup tasks https://github.com/openstack/openstack-ansible-plugins/blob/master/roles/db_setup/tasks/main.yml#L29 | 14:50 |
| jrosser | but there is a variable which allows that to be disabled | 14:50 |
| jrosser | you can run your playbook with `-e "_oslodb_setup_nolog=False"` | 14:52 |
| tuggernuts1 | oh sweet | 14:52 |
| tuggernuts1 | thanks for that | 14:52 |
| tuggernuts1 | I had issues with gnocci as well and applied https://review.opendev.org/c/openstack/openstack-ansible/+/887513 by adding gnocchi_git_install_branch: 6f35ea5413a9f78551d8193b8d2a6d77c49b6372 to user_variables.yml but then I started to get this https://pastebin.com/hHEzgaAT | 14:54 |
| tuggernuts1 | https://bugs.launchpad.net/openstack-ansible/+bug/2026828 | 14:56 |
| tuggernuts1 | that's the bug I think | 14:56 |
| jrosser | do you need gnocchi? | 15:00 |
| frickler | that issue should be fixed in gnocchi master since yesterday | 15:00 |
| tuggernuts1 | pretty much just trying to get the standard install to go but I probably don't need gnocchi | 15:03 |
| tuggernuts1 | how can I set this to pull latest? or just turn off this service? | 15:03 |
| jrosser | i would suggest not deploying any of the telemetry stack (celiometer / gnocchi etc etc) unless you actually need it | 15:03 |
| jrosser | "pull latest" is not really what you want | 15:04 |
| tuggernuts1 | I'm a bit confused on how to turn off or enable services honestly | 15:04 |
| tuggernuts1 | oh sorry, frickler said this gnocchi issue was fixed in master yesterday | 15:04 |
| tuggernuts1 | "latest" | 15:04 |
| jrosser | if you update openstack-ansible to the point release which includes the patch you linked to, then you will get the fix included | 15:05 |
| jrosser | the services that are deployed are determined by what yo do / do not put in your openstack_user_config.yml file | 15:05 |
| tuggernuts1 | ok I pulled 27.0.0 a few weeks ago is there a path to upgrade or just pull 27.x.x? | 15:05 |
| jrosser | there is now a 27.0.1 tag | 15:06 |
| tuggernuts1 | kk | 15:07 |
| jrosser | and there is a guide here for how to do a minor upgrade (i.e within a major release) https://docs.openstack.org/openstack-ansible/2023.1/admin/upgrades/minor-upgrades.html | 15:07 |
| jrosser | gnocchi "fixed in master" means fixed in the master version of gnocchi | 15:07 |
| tuggernuts1 | ok is there a doc on how to disable a service in the openstack_user_config? | 15:08 |
| tuggernuts1 | is it just removing the metering-compute_hosts stanza? | 15:09 |
| tuggernuts1 | and metrics_hosts I guess | 15:09 |
| jrosser | yes, but you will have to use scripts/inventory-manage.py to actually remove those hosts from the ansible inventory | 15:11 |
| tuggernuts1 | gotcha ok | 15:11 |
| jrosser | https://docs.openstack.org/openstack-ansible/2023.1/reference/inventory/manage-inventory.html | 15:11 |
| tuggernuts1 | thanks for the help | 15:11 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!