| *** N3l1x has quit IRC | 01:09 | |
| *** k_mouza has joined #openstack-containers | 01:31 | |
| *** k_mouza has quit IRC | 01:35 | |
| openstackgerrit | Feilong Wang proposed openstack/magnum master: [k8s] Support updating k8s cluster health status https://review.opendev.org/710384 | 02:19 |
|---|---|---|
| *** sapd1_x has joined #openstack-containers | 03:10 | |
| openstackgerrit | Merged openstack/magnum-ui master: Fix pyScss version in lower-constraints.txt https://review.opendev.org/716717 | 03:31 |
| *** ykarel|away is now known as ykarel | 04:33 | |
| *** udesale has joined #openstack-containers | 04:56 | |
| *** sapd1_x has quit IRC | 05:02 | |
| *** threestrands has joined #openstack-containers | 05:06 | |
| openstackgerrit | Feilong Wang proposed openstack/magnum master: [WIP][k8s] Update k8s dashboard verstion to v2.0.0 https://review.opendev.org/714021 | 05:09 |
| *** sapd1_x has joined #openstack-containers | 05:14 | |
| *** mnasiadka has quit IRC | 06:21 | |
| *** mnasiadka has joined #openstack-containers | 06:22 | |
| *** sapd1_x has quit IRC | 06:29 | |
| *** vishalmanchanda has joined #openstack-containers | 06:30 | |
| *** sapd1_x has joined #openstack-containers | 06:41 | |
| *** dioguerra has joined #openstack-containers | 07:14 | |
| *** sapd1_x has quit IRC | 07:39 | |
| *** threestrands has quit IRC | 08:01 | |
| *** ykarel is now known as ykarel|lunch | 08:08 | |
| cosmicsound | anyone here using latest fedora-coreos31 openstack ? if so what os tag you people use with magnum for k8s coreos or fedora-coreos ? | 09:25 |
| ttsiouts | strigazi, brtknr: thanks for the comments. I like the --add-labels approach | 09:27 |
| ttsiouts | this way we could have --remove-labels as well | 09:27 |
| ttsiouts | what do you think? | 09:28 |
| cosmicsound | I have added the label heat_container_agent_tag: train-stable-2 and my cluster still uses ussuri-dev | 09:37 |
| cosmicsound | I added also on template level also on cluster level nothing helps | 09:37 |
| cosmicsound | ANy ideas what tags are for template only and what tags are for k8s directly | 09:38 |
| *** ykarel|lunch is now known as ykarel | 09:42 | |
| *** k_mouza has joined #openstack-containers | 10:12 | |
| openstackgerrit | Bharat Kunwar proposed openstack/magnum-specs master: Fix CI and address all tox -e docs warnings https://review.opendev.org/716923 | 10:33 |
| openstackgerrit | Lingxian Kong proposed openstack/magnum master: [WIP][K8S] Delete all related load balancers before deleting cluster https://review.opendev.org/716930 | 10:51 |
| openstackgerrit | Bharat Kunwar proposed openstack/magnum-specs master: ci: Fix and address all tox -e docs warnings https://review.opendev.org/716923 | 10:53 |
| brtknr | ttsiouts: remove-labels sounds good :) | 10:54 |
| ttsiouts | brtknr: although I like it too, I am not sure that there is a valid use case for it. | 10:57 |
| ttsiouts | brtknr: meaning that removing a label should mean don't use it or use the default? | 10:58 |
| brtknr | yes, to be fair, i dont see myself using it much | 10:58 |
| ttsiouts | brtknr: and there is noone to ask for the default value.. from user perspective. | 10:58 |
| brtknr | ttsiouts: thats true, although another way to look at it is: a user is responsible for both template and the cluster | 11:15 |
| openstackgerrit | Bharat Kunwar proposed openstack/magnum-specs master: ci: Fix and address all tox -e docs warnings https://review.opendev.org/716923 | 11:18 |
| openstackgerrit | Theodoros Tsioutsias proposed openstack/magnum-specs master: [WIP] Magnum Labels Override https://review.opendev.org/716571 | 11:39 |
| *** udesale_ has joined #openstack-containers | 11:40 | |
| *** udesale has quit IRC | 11:43 | |
| *** rcernin has quit IRC | 11:54 | |
| *** ykarel is now known as ykarel|afk | 12:03 | |
| *** dioguerra has quit IRC | 12:10 | |
| *** ramishra has quit IRC | 12:10 | |
| *** ykarel|afk is now known as ykarel | 12:13 | |
| *** ramishra has joined #openstack-containers | 12:15 | |
| openstackgerrit | Theodoros Tsioutsias proposed openstack/magnum-specs master: Magnum Labels Override https://review.opendev.org/716571 | 12:28 |
| *** ykarel is now known as ykarel|afk | 12:33 | |
| ttsiouts | brtknr: I rebased the spec on top of the ci fix. I added the ussuri index too because the ci was failing without it | 12:51 |
| ttsiouts | dtomasgu: thanks for the comments. | 12:52 |
| ttsiouts | dtomasgu: do you want to add a command that will show only the labels of the cluster or nodegroup? | 12:52 |
| ttsiouts | dtomasgu: By get methods, I was referring to cluster/nodegroup show | 12:54 |
| *** ykarel|afk is now known as ykarel | 12:55 | |
| *** dioguerra has joined #openstack-containers | 12:56 | |
| dioguerra | i am here | 12:56 |
| ttsiouts | dioguerra: do you want to add a command that will show only the labels of the cluster or nodegroup? | 12:57 |
| ttsiouts | dioguerra: By get methods, I was referring to cluster/nodegroup show | 12:57 |
| ttsiouts | dioguerra: thanks for the comments. | 12:57 |
| dioguerra | ttsiouts: only cluster or nodegroup. otherwise if the user has multiple nodegroups with different 'user_labels' how are you going to merge them? | 12:58 |
| dioguerra | ttsiouts: alse, the default nodegroup inherits the labels from the cluster settings. | 12:59 |
| ttsiouts | dioguerra: hmm, I was thinking of just adding a new field in the show output | 12:59 |
| dioguerra | ttsiouts: as i understood you will add the 'user_labels' column under the Cluster and nodegroups table | 12:59 |
| ttsiouts | dioguerra: not adding a whole new command | 12:59 |
| ttsiouts | dioguerra: yes, user_labels will be added in both cluster and nodegroup | 13:00 |
| dioguerra | ttsiouts: so the filed in the output is based on the --labels input moving forward (if so that was not understood by me) | 13:00 |
| ttsiouts | dioguerra: so each show will contain the new field too | 13:00 |
| dioguerra | ^^ new field is 'user_labels' and it will also show labels? And what are the labels (template/Cluster only) or plus 'user_labels' | 13:03 |
| dioguerra | ? | 13:03 |
| ttsiouts | dioguerra: I was thinking something like this cluster.labels.update(cluster.user_labels) | 13:03 |
| ttsiouts | and for NGs the same | 13:03 |
| dioguerra | what will happen with the old labels field? | 13:04 |
| ttsiouts | the labels output should either be empty, or show the inherited labels | 13:05 |
| ttsiouts | not sure yet | 13:05 |
| dioguerra | because you can always compute the dependencies if you have the 'user_labels' input | 13:06 |
| dioguerra | s/input/column/ | 13:06 |
| ttsiouts | this is what we are trying to decide. | 13:06 |
| brtknr | so labels behaves exactly the same as it does now? | 13:07 |
| ttsiouts | IMHO, it feels better to be empty for clusters that use the --add/extra/new-labels | 13:07 |
| brtknr | and --user-labels overrides inherited labels? | 13:07 |
| ttsiouts | meaning that the conductor should fetch the labels from the cluster template before sending the request to heat | 13:08 |
| ttsiouts | I'm not sure yet though.. | 13:08 |
| dioguerra | I dont think 'user_labels' should interact with labels | 13:08 |
| dioguerra | unless it is the cluster_template labels obviously | 13:09 |
| dioguerra | then the conductor should fetch the cluster_template labels and apply 'new_labels' and then send request in the case of cluster creeate | 13:10 |
| dioguerra | in the case of NG create the conducter fetches the cluster labels and applies the 'new_labels' an sends the request | 13:10 |
| dioguerra | get for the cluster gets all used labels (so the computation of cluster_template + 'new_labels' | 13:11 |
| dioguerra | get ffor the NG gets all the used labels | 13:12 |
| dioguerra | you can also add a field for the 'new_labels if you want' or a --only_new_labels flag to return only the user defined labels | 13:12 |
| dioguerra | i think this makes sense because then you are going to add file_config (which is another column on the Cluster/NG table). Which behaves the same way as the labels now, taking precedence with: cluster_template->file->user_labels for the cluster create and cluster_labels->file->user_labels (im thinking on only allowing config file for cluster wide set-up tho) | 13:16 |
| *** ianychoi_ has joined #openstack-containers | 13:17 | |
| *** ianychoi has quit IRC | 13:20 | |
| ttsiouts | we need to be backwards compatible and not break the API | 13:22 |
| ttsiouts | so labels should continue working as they were. | 13:23 |
| *** ianychoi_ is now known as ianychoi | 13:23 | |
| ttsiouts | meaning that labels will continue being a way to tottaly override what was set in the level above | 13:24 |
| ttsiouts | user_labels should contain labels and values to be merged with the inherited labels | 13:24 |
| ttsiouts | does it make sense? | 13:25 |
| ttsiouts | let's continue in gerrit because I don't know if this is clear enough. | 13:27 |
| ttsiouts | I'll try to add all these in the spec so that we can comment on each part | 13:28 |
| brtknr | ttsiouts: what happens if a user specifies both --labels and --user-labels at the cluster scope | 13:28 |
| *** ykarel is now known as ykarel|afk | 13:29 | |
| brtknr | --labels specified at the time get merged with --user-labels? | 13:29 |
| ttsiouts | brtknr: https://review.opendev.org/#/c/716571/3/specs/ussuri/labels-override.rst@100 | 13:37 |
| ttsiouts | the api should throw an error | 13:37 |
| brtknr | ttsiouts: ah cool :) | 13:39 |
| brtknr | that solves that problem | 13:39 |
| brtknr | ttsiouts: i am still not happy with any of the flags | 13:54 |
| brtknr | --labels-override seems the best to me | 13:55 |
| brtknr | or --update-labels | 13:55 |
| brtknr | since this is equivalent to a dict update in python | 13:56 |
| *** vishalmanchanda has quit IRC | 13:58 | |
| *** dioguerra has quit IRC | 14:06 | |
| *** ykarel|afk is now known as ykarel | 14:28 | |
| *** ramishra has quit IRC | 14:35 | |
| *** ramishra has joined #openstack-containers | 14:38 | |
| *** ttsiouts has quit IRC | 14:49 | |
| *** ramishra has quit IRC | 14:58 | |
| *** ttsiouts has joined #openstack-containers | 15:08 | |
| cosmicsound | req-5c19c629-78c8-40ac-8ac6-568eec9fb3baattach_volume2 Apr 2020, 9:36 a.m.7096c304e64b4d5a87fa4d7cb8808042Error | 15:13 |
| cosmicsound | Is anyone getting such errors on mater of volume attached on master0k8s | 15:13 |
| *** ramishra has joined #openstack-containers | 15:14 | |
| *** ondrejburian has quit IRC | 15:19 | |
| brtknr | cosmicsound: no | 15:26 |
| brtknr | not seen that before | 15:27 |
| brtknr | can you explain the problem more? | 15:27 |
| brtknr | did you get a working cluster/ | 15:27 |
| cosmicsound | brtknr , this is the logs from a master vm and no i still did not managed to get magnum fully load a succesfull cluster | 15:33 |
| cosmicsound | like im cursed on this one | 15:33 |
| cosmicsound | It shows me this errors on machine view | 15:33 |
| cosmicsound | yet there is nothing wrong with the disk on the machine, the disk is there on right size | 15:33 |
| brtknr | cosmicsound: some info about what version you are using would be good, also example of the cluster template/cluster as i requested the other day | 15:34 |
| brtknr | there are so many configurations, it is hard to know what is wrong from 1 line error log message | 15:35 |
| cosmicsound | brtknr , true that, well we use kolla to deploy from source, so we have latest magnum 9.2.0 running | 15:40 |
| cosmicsound | I am not sure why no one makes a default template guide to know it works on a major release | 15:41 |
| cosmicsound | Its ridiculous that over 5 months not one cluster completed | 15:41 |
| cosmicsound | all failed in various reassons | 15:41 |
| *** ondrejburian has joined #openstack-containers | 15:42 | |
| *** ttsiouts has quit IRC | 15:46 | |
| *** ykarel is now known as ykarel|away | 15:47 | |
| cosmicsound | brtknr , i am confused on how tags are overwriten from template level to cluster level | 15:48 |
| cosmicsound | whats the main one to overrule the tags? for example in main template i gave heat_agent_tag = train-stable-2 | 15:48 |
| cosmicsound | and the. cluster still runs ussuri-dev | 15:48 |
| *** ttsiouts has joined #openstack-containers | 15:48 | |
| cosmicsound | http://paste.openstack.org/show/791549/ here are the logs from nova, related to the attach_volume failed | 16:00 |
| cosmicsound | https://mdb.uhlhost.net/uploads/670d2c24165cc042/image.png | 16:00 |
| cosmicsound | no where the same reqs were found in logs | 16:01 |
| cosmicsound | we use ceph backend | 16:01 |
| *** ttsiouts has quit IRC | 16:04 | |
| cosmicsound | so etcd does not run on my master:sudo service etcd status -l shows its not even existing | 16:09 |
| cosmicsound | fedora-atomic-29 here | 16:09 |
| cosmicsound | brtknr , this is my current template: http://paste.openstack.org/show/791552/ | 16:24 |
| *** udesale_ has quit IRC | 16:26 | |
| *** ttsiouts has joined #openstack-containers | 16:39 | |
| brtknr | You have space in the labels | 16:48 |
| cosmicsound | ok | 16:53 |
| *** ttsiouts has quit IRC | 16:54 | |
| cosmicsound | brtknr , seems also etcd is not running for some reasson inside the Fedora Atomic 29 | 17:00 |
| guilhermesp | cosmicsound: did you try ( if you are running mangum 9.2.0 ) to use labels as use_podman=yes and etcd_tag=3.4.3? not sure if i missed in the pastes but which kube_tag are you using? | 17:03 |
| cosmicsound | I did not set any. kube_tag | 17:04 |
| cosmicsound | was the default on 1.5.7 | 17:04 |
| cosmicsound | these labels needs to be on template or on cluster? | 17:05 |
| guilhermesp | depending on the k8s version you're willing to use with train 9.2.0 yeah, that's how it works for me to support v1.17/v1.16 | 17:06 |
| cosmicsound | Labels | 17:06 |
| cosmicsound | cloud_provider_enabled=true | 17:06 |
| cosmicsound | heat_container_agent_tag=train-stable-2 | 17:06 |
| cosmicsound | etcd_volume_size=20 | 17:06 |
| cosmicsound | use_podman=yes | 17:06 |
| cosmicsound | etcd_tag=3.4.3 | 17:06 |
| cosmicsound | We have these tags on template now | 17:06 |
| cosmicsound | sry. the long paste | 17:06 |
| cosmicsound | issue is etcd service is not running at all | 17:07 |
| cosmicsound | guilhermesp , what image you use? | 17:07 |
| cosmicsound | fedora atomic or coreos | 17:07 |
| guilhermesp | I've been using atomic, but we are planning to migrate our cloud to apply some fixes that allow us to use fedora-coreos | 17:07 |
| guilhermesp | as fedora atomic is EOL | 17:08 |
| cosmicsound | true that | 17:08 |
| guilhermesp | i had some issues of etcd not running, but that has to do with which etcd_tag you are using for which kubernetes version | 17:08 |
| cosmicsound | ok | 17:08 |
| guilhermesp | so yeah, in train 9.2.0 i was able to run 1.16 cleanly and v1.17 | 17:08 |
| cosmicsound | nice nice | 17:08 |
| cosmicsound | can you give me some peak on your template tags used? | 17:09 |
| guilhermesp | but v1.17 on that tag is not passing on conformance tests so, there is a fix but it that is still not backported i guess ( not sure if is going to be tbh ) | 17:09 |
| guilhermesp | cosmicsound: that's the v.1.16 that I use cleanly in train 9.2.0 http://paste.openstack.org/show/791555/ | 17:11 |
| guilhermesp | also passes k8s conformance tests | 17:11 |
| guilhermesp | i'm using that cloud-provider coz we has issues to attach cinder volumes with v1.14.0 | 17:11 |
| guilhermesp | but yeah, that will be the last template that I will use atomic images hopefully :) | 17:12 |
| guilhermesp | and we are going to be providing support only for the two recent k8s versions so we are no longer going to use prior to v1.16 | 17:13 |
| guilhermesp | one thing I advice you is to follow the magnum team weekly meeting | 17:13 |
| cosmicsound | guilhermesp , behind ceph? | 17:13 |
| guilhermesp | yes we have all stuff behind ceph | 17:13 |
| cosmicsound | guilhermesp , i am here in past 5 months :D | 17:14 |
| cosmicsound | every day | 17:14 |
| cosmicsound | guilhermesp , use kolla by chance? | 17:14 |
| guilhermesp | cosmicsound: lol | 17:14 |
| guilhermesp | nops | 17:14 |
| *** ttsiouts has joined #openstack-containers | 17:14 | |
| guilhermesp | we use OSA | 17:14 |
| cosmicsound | right | 17:14 |
| guilhermesp | but yeah, i think if you are sourcing the sha of train 9.2.0 to build your packages and stuff | 17:15 |
| cosmicsound | so must I use also rbd | 17:15 |
| cosmicsound | since i use also ceph | 17:15 |
| guilhermesp | i think those labels would make you able to run v1.16 | 17:15 |
| guilhermesp | yeah i'm not sure why heat does'nt complain, unless it has a default value that is equal for your cinder volume types | 17:16 |
| cosmicsound | how come you choose calico over flannel? | 17:16 |
| guilhermesp | s/beat/heat | 17:16 |
| cosmicsound | got issues with flannel? | 17:16 |
| guilhermesp | for me v1.16 runs both calico and flannel ok | 17:16 |
| cosmicsound | right | 17:16 |
| guilhermesp | but yeah so my journey now is to provide clean support to v1.17 and v1.18 so, I will be a bit more active around | 17:17 |
| *** KeithMnemonic has joined #openstack-containers | 17:28 | |
| *** k_mouza has quit IRC | 17:46 | |
| *** k_mouza has joined #openstack-containers | 18:13 | |
| brtknr | cosmicsound: | 18:14 |
| *** k_mouza has quit IRC | 18:14 | |
| brtknr | your labels only shows this: | 18:14 |
| brtknr | labels: | 18:14 |
| brtknr | ' etcd_volume_size': '20' | 18:14 |
| brtknr | ' heat_container_agent_tag': train-stable-2 | 18:14 |
| brtknr | cloud_provider_enabled: 'true' | 18:15 |
| brtknr | you have not set etcd_tag | 18:15 |
| brtknr | also for some reason, your label is parsed as ' etcd_volume_size' | 18:15 |
| brtknr | there should be no space before etcd_volume_size | 18:15 |
| brtknr | i would check the script you are using to launch your cluster | 18:15 |
| brtknr | do you have it on a public repository somewhere? | 18:15 |
| cosmicsound | brtknr , i set it on cluster level | 18:40 |
| cosmicsound | and last when i set it on template level was not working | 18:40 |
| cosmicsound | brtknr , not sure how you mean check the script to launch the cluster. | 18:41 |
| cosmicsound | I follow manual commands, create a image from atomic fedora, create the template manually via cli and run the cluster. | 18:41 |
| cosmicsound | Resource CREATE failed: Error: resources.kube_masters.resources[0].resources.master_config_deployment: Deployment to server failed: deploy_status_code: Deployment exited with non-zero status code: 1 | 19:01 |
| brtknr | cosmicsound: I need to see all the labels | 19:43 |
| brtknr | both cluster template and cluster | 19:43 |
| brtknr | If you set labels at cluster level, the cluster template labels have no effect | 19:44 |
| *** ttsiouts has quit IRC | 19:56 | |
| openstackgerrit | Spyros Trigazis proposed openstack/magnum master: WIP Support calico v3.3.6 https://review.opendev.org/717116 | 20:02 |
| openstackgerrit | Spyros Trigazis proposed openstack/magnum master: Support calico v3.3.6 https://review.opendev.org/717116 | 20:21 |
| openstackgerrit | Spyros Trigazis proposed openstack/magnum master: Support calico v3.3.6 https://review.opendev.org/717116 | 20:24 |
| *** flwang1 has joined #openstack-containers | 20:43 | |
| flwang1 | brtknr: ping | 20:45 |
| *** ttsiouts has joined #openstack-containers | 20:45 | |
| flwang1 | re the health status update patch | 20:45 |
| *** ttsiouts_ has joined #openstack-containers | 20:46 | |
| *** ttsiouts has quit IRC | 20:49 | |
| cosmicsound | brtknr , http://paste.openstack.org/show/791562/ | 21:50 |
| *** k_mouza has joined #openstack-containers | 22:15 | |
| *** k_mouza has quit IRC | 22:19 | |
| *** rcernin has joined #openstack-containers | 22:22 | |
| *** ttsiouts_ has quit IRC | 22:38 | |
| *** openstackstatus has quit IRC | 22:56 | |
| *** openstack has joined #openstack-containers | 22:59 | |
| *** ChanServ sets mode: +o openstack | 22:59 | |
| *** ttsiouts has joined #openstack-containers | 23:16 | |
| *** ttsiouts has quit IRC | 23:21 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!
