| *** zzzeek has quit IRC | 01:39 | |
| *** zzzeek has joined #openstack-kolla | 01:41 | |
| *** ricolin_ has joined #openstack-kolla | 01:42 | |
| *** vishalmanchanda has joined #openstack-kolla | 03:59 | |
| *** zzzeek has quit IRC | 04:34 | |
| *** zzzeek has joined #openstack-kolla | 04:37 | |
| *** shyamb has joined #openstack-kolla | 04:39 | |
| *** shyamb has quit IRC | 04:53 | |
| *** ricolin_ has quit IRC | 05:23 | |
| *** ricolin has joined #openstack-kolla | 05:24 | |
| *** zzzeek has quit IRC | 05:33 | |
| *** zzzeek has joined #openstack-kolla | 05:36 | |
| *** zzzeek has quit IRC | 05:57 | |
| *** zzzeek has joined #openstack-kolla | 05:59 | |
| *** cah_link has joined #openstack-kolla | 06:29 | |
| *** k_mouza has joined #openstack-kolla | 06:40 | |
| *** k_mouza has quit IRC | 06:45 | |
| *** cah_link has quit IRC | 07:07 | |
| *** cah_link has joined #openstack-kolla | 07:19 | |
| *** ricolin has quit IRC | 07:32 | |
| *** hjensas has joined #openstack-kolla | 07:53 | |
| *** rpittau|afk is now known as rpittau | 08:00 | |
| *** maharg101 has joined #openstack-kolla | 08:08 | |
| *** jbadiapa has joined #openstack-kolla | 08:09 | |
| *** skramaja has joined #openstack-kolla | 08:09 | |
| *** kevko has joined #openstack-kolla | 08:20 | |
| *** kevko_ has joined #openstack-kolla | 08:22 | |
| *** kevko has quit IRC | 08:23 | |
| *** bengates has joined #openstack-kolla | 08:23 | |
| *** jbadiapa has quit IRC | 08:27 | |
| *** bengates has quit IRC | 08:28 | |
| *** bengates has joined #openstack-kolla | 08:29 | |
| openstackgerrit | Gaël THEROND proposed openstack/kolla-ansible master: Add missing elasticsearch cloudkitty storage and prometheus collector backend support. https://review.opendev.org/c/openstack/kolla-ansible/+/752965 | 08:29 |
|---|---|---|
| *** eliaswimmer has joined #openstack-kolla | 08:34 | |
| *** Fl1nt has joined #openstack-kolla | 08:35 | |
| Fl1nt | aaaah far better ^^ | 08:35 |
| Fl1nt | So! Good morning team! | 08:35 |
| Fl1nt | mgoddard, I've checked a bit what's required to get OIDC and SAML2 Federated authentication, I think if we merge the current OIDC patch with its current shape, it will conflict on a lot of things with the incoming SAML2 support. | 08:38 |
| Fl1nt | Especially on keystone/mellon WSGI apache config | 08:38 |
| Fl1nt | there is a way to get both OIDC/SAML2 supported at the same time as we currently do it. | 08:39 |
| mnasiadka | Fl1nt: so comment on the change which parts are conflicting? | 08:39 |
| Fl1nt | mnasiadka, already done it. | 08:39 |
| Fl1nt | but the assignee didn't take care of it. | 08:39 |
| mnasiadka | I'm feeling the OIDC patch will not make it in Wallaby. | 08:40 |
| Fl1nt | for instance, the mellon endpoint used within keystone apache's virtualhost shouldn't contain any specific OIDC reference but make use of wildcard/joker chars in order to let keystone route the OIDC or SAML2 request to the appropriate module | 08:41 |
| Fl1nt | accordingly | 08:41 |
| Fl1nt | as it's made to manage multiple authentication protocol in the same time. | 08:41 |
| Fl1nt | With the current shape, only OIDC or only SAML2 will be supported, but multiple platforms support both of them at once as they've got sparse authentication sources. | 08:42 |
| *** jbadiapa has joined #openstack-kolla | 08:54 | |
| kevko_ | morning :) ! | 08:55 |
| Fl1nt | Hi kevko_ ! | 08:57 |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible stable/victoria: docs: Improve multinode Docker registry setup https://review.opendev.org/c/openstack/kolla-ansible/+/774398 | 08:57 |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible stable/ussuri: docs: Improve multinode Docker registry setup https://review.opendev.org/c/openstack/kolla-ansible/+/774399 | 08:57 |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible stable/train: docs: Improve multinode Docker registry setup https://review.opendev.org/c/openstack/kolla-ansible/+/774400 | 08:58 |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible stable/victoria: docs: improve external Ceph docs https://review.opendev.org/c/openstack/kolla-ansible/+/774401 | 08:58 |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible stable/ussuri: docs: improve external Ceph docs https://review.opendev.org/c/openstack/kolla-ansible/+/774402 | 08:58 |
| *** dougsz has joined #openstack-kolla | 08:59 | |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible master: docs: Add information on tuning Ansible https://review.opendev.org/c/openstack/kolla-ansible/+/758205 | 09:02 |
| Fl1nt | mgoddard, cool bunch of additional docs, especially the ansible 'tunning' one, I've found out on friday that kayobe had that and was interesting ^^ BTW: Kayobe contain other docs that are interesting for kolla but I don't know if duplicating them is worthy instead of just cross pointing them maybe? | 09:06 |
| mgoddard | Fl1nt: thanks. If there are particular kayobe docs you'd like to see in kolla we can look at moving or referencing them | 09:07 |
| mgoddard | Fl1nt: IMO OIDC is ready to merge. If we have to change it to support SAML then so be it | 09:08 |
| hrw | morning | 09:09 |
| Fl1nt | Ok, let it go with OIDC, we'll deal with changes when I'll do the SAML integration. I'll then do a more generic patch to support OIDC and SAML2 at once. | 09:09 |
| Fl1nt | mgoddard, there are still two little things to fix with OIDC, I've added comment about that. | 09:10 |
| Fl1nt | Hi hrw how are you? | 09:11 |
| hrw | hoping for lack of fosdem flu | 09:11 |
| Fl1nt | ah ah ah I guess it's a new challenge to embrace, new slogan: "Test your immune system each time you get a toe outside!" | 09:13 |
| hrw | ;D | 09:20 |
| hrw | mgoddard: https://review.opendev.org/c/openstack/kolla/+/774351 fixes Debian builds | 09:26 |
| hrw | mnasiadka: ^^ | 09:27 |
| Fl1nt | All right, so, next patch is either prometheus support for external CEPH MGR Exporter (native mgr prometheus exporter) or Keystone federation using SAML2, what are your calls on that team? | 09:27 |
| mgoddard | Fl1nt: we already support prometheus Ceph mgr exporter | 09:30 |
| mgoddard | Fl1nt: so SAML :) | 09:30 |
| Fl1nt | on external ceph ? | 09:30 |
| mgoddard | enable_prometheus_ceph_mgr_exporter | 09:30 |
| mgoddard | prometheus_ceph_mgr_exporter_endpoints | 09:30 |
| Fl1nt | Oh ok, my bad, I though it was just aiming for kolla installed ceph clusters. | 09:31 |
| mgoddard | Fl1nt: perhaps you could document it, then the next person will know about it | 09:31 |
| Fl1nt | yep definitely | 09:31 |
| Fl1nt | another thing that I'd like to introduce, is to use https://docs.ansible.com/ansible/latest/collections/openstack/cloud/ with kolla_toolbox as much as possible when we can, like for SAML2/OIDC domains creation for instance | 09:32 |
| Fl1nt | that would simplify and shorten a bit things such as: https://review.opendev.org/c/openstack/kolla-ansible/+/695432/54/ansible/roles/keystone/tasks/register_identity_providers.yml | 09:33 |
| *** Fl1nt has quit IRC | 09:34 | |
| *** Fl1nt has joined #openstack-kolla | 09:35 | |
| Fl1nt | re, sorry, mistakenly closed the wrong window ^^ | 09:36 |
| *** e0ne has joined #openstack-kolla | 09:36 | |
| mgoddard | Fl1nt: it requires ansible 2.10 IIUC | 09:47 |
| Fl1nt | for the collection yes, but there is the os_<module> for under 2.10 series | 09:48 |
| *** k_mouza has joined #openstack-kolla | 09:59 | |
| *** kukacz has quit IRC | 10:10 | |
| *** e0ne has quit IRC | 10:17 | |
| *** kukacz has joined #openstack-kolla | 10:18 | |
| *** e0ne has joined #openstack-kolla | 10:18 | |
| *** e0ne_ has joined #openstack-kolla | 10:21 | |
| *** e0ne has quit IRC | 10:21 | |
| *** e0ne_ has quit IRC | 10:23 | |
| *** e0ne has joined #openstack-kolla | 10:23 | |
| *** e0ne has quit IRC | 10:25 | |
| *** e0ne has joined #openstack-kolla | 10:25 | |
| mgoddard | Fl1nt: I did look for suitable modules. There was one, but it was only in the collection, not os_* | 10:28 |
| Fl1nt | Yes endpoint is in collection, but once we will have switched to 2.10 we will be able to use endpoint and domain/project/federation/network/identity_* etc | 10:34 |
| Fl1nt | all right, so let's go with SAML2 federation then ^^ | 10:37 |
| mnasiadka | mgoddard: what is stopping us from bumping toolbox to 2.10? | 10:46 |
| mgoddard | mnasiadka: pass | 10:47 |
| mgoddard | you might read that either as 'I don't know' or zuul pass :) | 10:47 |
| mnasiadka | haha | 10:47 |
| mnasiadka | well, 2.8 is already eol | 10:47 |
| mgoddard | I thought someone proposed collections in toolbox before | 10:48 |
| mgoddard | not sure how far it got | 10:48 |
| kevko_ | yoctozepto: proxysql upgraded by kolla-ansible in production ! :) load fell down about 40 percent because we moved neutron and nova databases to another cluster ..an workinf flawlessly | 10:48 |
| mnasiadka | mgoddard: I did, but I got very much swamped into $job | 10:48 |
| mnasiadka | I wanted to use some additional modules I think | 10:49 |
| mnasiadka | For Ansible version 2.10 or later, the major release is maintained for one release cycle. When the next release comes out (for example, 2.11), the older release (2.10 in this example) is no longer maintained. | 10:49 |
| mnasiadka | what the heck? | 10:49 |
| mnasiadka | something changed? | 10:49 |
| *** kevko_ has quit IRC | 10:49 | |
| *** kevko has joined #openstack-kolla | 10:49 | |
| mgoddard | mnasiadka: maybe they will go for longer majors now they have a smaller core | 10:50 |
| mgoddard | mnasiadka: and of course they mean minor, not major :) | 10:50 |
| Fl1nt | kevko, how is your proxysql nodes balancing strategy? HA with Active/Active or Active/Passive ? | 10:50 |
| Fl1nt | Do you use USER hash to balance within shards or another strategy? | 10:51 |
| kevko | Fl1nt: active-passive (for now) we haven't tested it in active-active ... | 10:52 |
| kevko | Fl1nt: but it is about small change in proxysql | 10:52 |
| kevko | Fl1nt: reason to not switch to active-active was that openstack can have problem with it | 10:53 |
| Fl1nt | mgoddard, not sure they mean, minor, RH is trying to accelerate release cycle everywhere, they even proposed openstack to be every 3 months instead of the current pace which is actually ridiculous as planning a platform upgrade, even with current tools and upgrade technics require some preparation works that are more than 3 months depending on your infrastructure perimeter. | 10:53 |
| Fl1nt | kevko, are you refering to this: https://github.com/sysown/proxysql/issues/1873 ? | 10:53 |
| Fl1nt | it's too bad that the openstack foundation never decided to migrate the SQL structure to either NoSQL or take profit of proxySQL solution. I mean, using NoSQL schemas or SQL with ProxySQL would let them have a really efficient and quick solution for large multi-contient clusters | 10:56 |
| mgoddard | kevko: there is a thread about scaling on openstack-discuss: http://lists.openstack.org/pipermail/openstack-discuss/2021-February/020174.html | 10:57 |
| mgoddard | kevko: people talk about clusters of 1k - 1.5k nodes without DB issues | 10:57 |
| kevko | Fl1nt: we can't meet this issue ...as we are using *only* one server always as active ..second as passive (as it is now in haproxy) issue which you are referencing to is not problem proxysql ..but problem of openstack (i think) ..as openstack's read write solution is buggy ...or it is not ready to use read/write | 10:57 |
| mgoddard | kevko: you said you hit issues at 100 nodes. What are you doing differently? Lots of small resources? | 10:58 |
| Fl1nt | kevko, I know, I asked because you told me, we can think of using active/active once some nasty bugs would have been solved, so I though about this one, but it's just an example. | 10:58 |
| *** k_mouza has quit IRC | 11:00 | |
| kevko | mgoddard: sorry, i don't know what you are asking ..do you asking me why we had performance problem against DB or what ? | 11:01 |
| hrw | https://marcin.juszkiewicz.com.pl/2021/02/08/fosdem-2021-was-the-best-online-event-ever/ | 11:01 |
| mgoddard | kevko: basically, yes | 11:02 |
| kevko | mgoddard: maybe lot of requests to OS API from k8s ( there are about 500 clusters in Openstack) | 11:03 |
| mgoddard | kevko: could be, I've heard k8s can give OS a hammering | 11:04 |
| mgoddard | 500 clusters on 100 nodes? | 11:04 |
| Fl1nt | mgoddard, I worked with OVH closely lately, their infrastructure is garbage, no CELLs, a lot of in house buggy patches, etc. It's not really their IT guys fault but rather their market that generate such issue. | 11:05 |
| *** e0ne has quit IRC | 11:05 | |
| kevko | mgoddard: well, it's small instances on computes .. | 11:05 |
| mgoddard | Fl1nt: I'm sure they would appreciate you saying that | 11:05 |
| mgoddard | kevko: what's the application? | 11:06 |
| *** k_mouza has joined #openstack-kolla | 11:06 | |
| Fl1nt | kevko, how large is your clusters on average? Because that's 5 cluster per nodes, depending on your nodes size and clusters size it means what? 3/5/10 instances per clusters? that would mean 50 instances per nodes, that's not really heavy. | 11:06 |
| kevko | mgoddard: again i don't know what are u asking :D ... I really don't know details about their applications running ...we are just delivering and administering their openstack installation | 11:07 |
| Fl1nt | however yes k8s is heavily hammering on APIs. | 11:07 |
| kevko | well, honestly it is very hard to switch off some api ..because after few minutes they are reclustering because of API ... | 11:08 |
| kevko | it is very sensitive openstack environment | 11:08 |
| Fl1nt | mgoddard, oh don't worry, they already know it ^^, I was just telling it because of the mail discussion where OVH told everyone, we hardcode 1500 hypervisors per region, but they also tell everyone they don't use cell to their discharge. | 11:09 |
| Fl1nt | kevko, do you manage your k8s through magnum or are they native/rancher clusters hosted on nova? | 11:09 |
| hrw | mgoddard, mnasiadka, yoctozepto: https://review.opendev.org/c/openstack/kolla/+/774351 - review to unblock Debian in master | 11:10 |
| kevko | Fl1nt: honestly, magnum is not production ready (even if they are telling that is :D in documentation )... no if my informations is correct they are using kubespray for deploy k8s | 11:11 |
| *** e0ne has joined #openstack-kolla | 11:11 | |
| *** e0ne has quit IRC | 11:12 | |
| kevko | Fl1nt: I've tested magnum for quite long time ago ..and it is very nasty ... I don't believe it can be maintained flawlessly if installation is based on BASH scripts .. | 11:13 |
| Fl1nt | I don't know, we use magnum in prod heavily and faced only minor issues, I'm geniunely curious to know more about what were your issues? | 11:13 |
| kevko | Fl1nt: i'm not saying that magnum is BAD ... I can imagine (for example) to use magnum ..but replace fedora-atomic driver with (for example) kubespray-driver ... | 11:14 |
| kevko | Fl1nt: which driver do u use ? | 11:16 |
| kevko | fedora-atomic ? | 11:16 |
| Fl1nt | fedora-atomic now a day, but CoreOS for a long time, so we still have a bunch of k8s cluster using CoreOS images. | 11:16 |
| Fl1nt | remaining bash scripts could easily be replaced either with python equivalent tho. | 11:18 |
| kevko | Fl1nt: I know, but you have to write that script :S | 11:20 |
| kevko | *:D | 11:20 |
| Fl1nt | kevko, I'll have a look at how much bash scripts remain because last time we installed magnum and created the images etc we didn't faced a lot of (probably more on images than on magnum itself). | 11:23 |
| kevko | Fl1nt: I would really like to have kubespray driver in magnum | 11:26 |
| kevko | Fl1nt: it could be very cool | 11:26 |
| kevko | Fl1nt: Well maintained project, easy, ansible | 11:26 |
| Fl1nt | For now I've few patches to work on for Kolla/Kolla-ansible our next round of openstack core project contribution will be next year or at least end of 2021. | 11:27 |
| Fl1nt | I'll probably put some ressources on magnum at that time. | 11:27 |
| kevko | Fl1nt: back to DB active-active ... I think openstack has problem especcially with this ( Openstack sometimes need READ UNCOMMITTED ) => https://mydbops.wordpress.com/2018/06/22/back-to-basics-isolation-levels-in-mysql/ | 11:27 |
| mgoddard | Fl1nt: what happened to podman? | 11:31 |
| Fl1nt | what do you mean? | 11:31 |
| mgoddard | Fl1nt: I think you said you would be able to work on it early in 2021 | 11:32 |
| Fl1nt | kevko, READ UNCOMMITED are marginal now a day with openstack, as long as you're operating over "standard" big enough servers. | 11:32 |
| kevko | Fl1nt: do u have patches somewhere in review ? or just locally ? could u share ? | 11:32 |
| Fl1nt | mgoddard, yep, working on it, we're currently migrating to CentOS Stream, once that migration achieve we will try to get a podman based staging environment. | 11:33 |
| Fl1nt | Nothing for now still a WIP, I'll poke you as soon as we'll have more than just assumption that it can work and where it can break ^^ | 11:34 |
| mgoddard | Fl1nt: I made a small PoC: https://github.com/stackhpc/kolla-ansible/commit/e44d4b028e3aa24955dd12271783287ae43a5603 | 11:35 |
| Fl1nt | oh! cool! Going to look at it. | 11:35 |
| mgoddard | Fl1nt: the systemd unit formatting was broken, didn't really look into it | 11:36 |
| Fl1nt | I see how you did test it, that's clever, going with service focused strategy rather than globally. | 11:39 |
| mgoddard | Fl1nt: I would suggest continuing that approach upstream, finding a pattern that works for the community, before implementing for all services | 11:44 |
| Fl1nt | yep | 11:45 |
| *** cah_link has quit IRC | 11:50 | |
| *** Luzi has joined #openstack-kolla | 11:55 | |
| openstackgerrit | Merged openstack/kolla stable/victoria: Fix Mistral source images to respect upper-constraints https://review.opendev.org/c/openstack/kolla/+/773012 | 12:29 |
| openstackgerrit | Merged openstack/kolla stable/ussuri: Fix Mistral source images to respect upper-constraints https://review.opendev.org/c/openstack/kolla/+/773013 | 12:33 |
| *** gfidente has joined #openstack-kolla | 12:56 | |
| kevko | guys, let's say that i want to change wsgi daemons but only for keystone .. it is openstack_service_workers variable and it is defined in all.yml group vars ...if i add keystone:vars to inventory ..and try to rewrite there ..should this be workable ? because it is not ..OR i am doing something wrong | 13:00 |
| *** alanmeadows has quit IRC | 13:26 | |
| *** stackedsax has quit IRC | 13:26 | |
| *** gouthamr has quit IRC | 13:26 | |
| *** snbuback has quit IRC | 13:27 | |
| *** emccormick has quit IRC | 13:27 | |
| *** jopdorp has quit IRC | 13:27 | |
| *** sri_ has quit IRC | 13:27 | |
| *** johnsom has quit IRC | 13:27 | |
| *** cz3 has quit IRC | 13:27 | |
| *** gmann has quit IRC | 13:27 | |
| *** vishalmanchanda has quit IRC | 13:27 | |
| *** zul has quit IRC | 13:27 | |
| *** parallax has quit IRC | 13:27 | |
| *** PrinzElvis has quit IRC | 13:27 | |
| *** alanmeadows has joined #openstack-kolla | 13:28 | |
| *** johnsom has joined #openstack-kolla | 13:28 | |
| *** cz3 has joined #openstack-kolla | 13:29 | |
| *** stackedsax has joined #openstack-kolla | 13:29 | |
| *** jopdorp has joined #openstack-kolla | 13:29 | |
| *** emccormick has joined #openstack-kolla | 13:29 | |
| *** sri_ has joined #openstack-kolla | 13:29 | |
| *** vishalmanchanda has joined #openstack-kolla | 13:29 | |
| *** gouthamr has joined #openstack-kolla | 13:29 | |
| *** snbuback has joined #openstack-kolla | 13:31 | |
| *** gmann has joined #openstack-kolla | 13:31 | |
| *** parallax has joined #openstack-kolla | 13:31 | |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible master: service-ks-register: ensure all user roles exist https://review.opendev.org/c/openstack/kolla-ansible/+/774451 | 13:31 |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible master: Support assigning additional roles to the nova user https://review.opendev.org/c/openstack/kolla-ansible/+/774452 | 13:32 |
| *** zul has joined #openstack-kolla | 13:33 | |
| hrw | mgoddard: looks like we need to backports sensu down to stein. | 13:44 |
| mgoddard | hrw: ok | 13:45 |
| kevko | could you please advise me ? | 13:47 |
| kevko | with that groups wars please ? | 13:48 |
| Fl1nt | kevko, looking | 13:48 |
| yoctozepto | hrw, mgoddard: is not tripleo building sensu though? | 13:49 |
| kevko | Fl1nt: because I think openstack_service_workers variable was meant ..per service .. and you really need to be able define workers per project ..because not every service has same load, right ... so I wondered that it should be able to define in keystone (or service) group vars and override it ... but it seems it's not possible because all is parent of everything | 13:50 |
| yoctozepto | (I somehow feel I have asked this question before) | 13:50 |
| yoctozepto | (sensu creep) | 13:50 |
| Fl1nt | kevko, you can override it using group_vars/host_vars | 13:50 |
| *** PrinzElvis has joined #openstack-kolla | 13:51 | |
| Fl1nt | or you can simply replace each <service>_wsgi_processes within your globals.yml or service. | 13:52 |
| *** Luzi has quit IRC | 13:52 | |
| hrw | yoctozepto: does not change anything. for ubuntu we use sensu repos which are no more | 13:54 |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible master: Support assigning additional roles to service users https://review.opendev.org/c/openstack/kolla-ansible/+/774452 | 13:56 |
| kevko | Fl1nt: well, group_vars not working ... | 13:57 |
| kevko | Fl1nt: I suppose that host_var will work | 13:57 |
| yoctozepto | hrw: ack, then drop only ubuntu, makes sense | 13:58 |
| Fl1nt | inventory group_vars/all is 3 in order for precedence, meaning really low priority. inventory group_vars/* is 6 and playbook group_vars/* is 8 it SHOULD work | 14:00 |
| Fl1nt | with something like group_vars/keystone.yml | 14:00 |
| Fl1nt | or keystone-<service>.yml | 14:01 |
| kevko | Fl1nt: well, i defined [keystone:vars] for test (didn't want to create groups_vars/keystone.yml -> because in-fact it is "changing" code ) | 14:02 |
| kevko | *in inventory | 14:03 |
| Fl1nt | it won't work | 14:03 |
| mgoddard | kevko: you need to use a custom config file, not ansible variable | 14:04 |
| Fl1nt | I can't remember why but it doesn't work even if it's supposed to work, but if you put your vars inline and duplicate them within your inventory instead of having the array reference notation it will | 14:04 |
| mgoddard | ansible variables are scoped to hosts, not services | 14:04 |
| Fl1nt | mgoddard, yes but you can use this notation on regular ansible relatively simple playbooks where you declare groups variable within the inventory as shown by kevko. | 14:05 |
| Fl1nt | however | 14:06 |
| Fl1nt | you're right on our case you can do that by having inline per host vars | 14:06 |
| Fl1nt | as someone else had this issue last year | 14:06 |
| mgoddard | you can't use ansible variables if you want different services on the same host to use a different value | 14:07 |
| Fl1nt | but ansible doc told that having group vars at inventory level is possible with array reference notation as long as you use this reference at host level. | 14:07 |
| mgoddard | in that case you need to use custom config files | 14:07 |
| Fl1nt | yes, hence why I suggested kevko to use group_vars/keystone.yml or group_vars/keystone_<service>.yml | 14:08 |
| mgoddard | Fl1nt: it doesn't work | 14:08 |
| mgoddard | ansible doesn't understand services, only hosts | 14:08 |
| mgoddard | you need /etc/kolla/config/keystone.conf | 14:09 |
| Fl1nt | It understand groups | 14:09 |
| kevko | mgoddard: why ? ansible should understand groups | 14:09 |
| mgoddard | Fl1nt: yes, but groups are groups of hosts | 14:09 |
| Fl1nt | yes | 14:09 |
| mnasiadka | mgoddard: rather wsgi-keystone.conf - or however it's called ;) | 14:09 |
| mnasiadka | (if we're talking openstack_workers) | 14:09 |
| mgoddard | yeah I guess | 14:09 |
| mgoddard | maybe I don't understand the problem | 14:09 |
| Fl1nt | so when let say you declare group_vars/keystone.yml it will apply override vars of all host in keystone group that use controller group as parent and so node01/02 for instance, but if you add glance.yml it will do the same but per service. | 14:10 |
| mgoddard | nope | 14:11 |
| Fl1nt | but on same group of host | 14:11 |
| mgoddard | not unless you have every service on a separate host | 14:11 |
| Fl1nt | ooooooh I know what you mean. | 14:11 |
| Fl1nt | ok | 14:11 |
| kevko | yeah, i got it also | 14:12 |
| mgoddard | :) | 14:13 |
| kevko | what a pity :( | 14:13 |
| Fl1nt | You're right, if your hosts are all grouped above controllers etc as services use those groups as parent :D | 14:13 |
| Fl1nt | BUT | 14:13 |
| Fl1nt | do you have everything located as control plan and compute plan kevko ? | 14:14 |
| kevko | yeah | 14:14 |
| Fl1nt | arff | 14:19 |
| openstackgerrit | Merged openstack/kolla stable/victoria: Horizon: gentler `-o nounset` handling https://review.opendev.org/c/openstack/kolla/+/773196 | 14:23 |
| openstackgerrit | Merged openstack/kolla master: [docs] Add templates and examples of renos https://review.opendev.org/c/openstack/kolla/+/773261 | 14:25 |
| *** tbarron is now known as tbarron|out | 15:22 | |
| *** k_mouza has quit IRC | 15:25 | |
| *** k_mouza has joined #openstack-kolla | 15:26 | |
| *** vishalmanchanda has quit IRC | 15:45 | |
| *** hjensas is now known as hjensas|afk | 15:46 | |
| *** skramaja has quit IRC | 15:47 | |
| *** k_mouza_ has joined #openstack-kolla | 16:01 | |
| *** cah_link has joined #openstack-kolla | 16:04 | |
| *** k_mouza has quit IRC | 16:05 | |
| *** e0ne has joined #openstack-kolla | 16:06 | |
| *** k_mouza_ has quit IRC | 16:08 | |
| *** k_mouza has joined #openstack-kolla | 16:08 | |
| *** Fl1nt has quit IRC | 16:13 | |
| *** e0ne has quit IRC | 16:49 | |
| *** k_mouza has quit IRC | 16:52 | |
| *** k_mouza has joined #openstack-kolla | 16:53 | |
| *** jonaspaulo has joined #openstack-kolla | 17:05 | |
| *** rpittau is now known as rpittau|afk | 17:05 | |
| *** cah_link has quit IRC | 17:12 | |
| mgoddard | kolla-ansible CI broken by python cryptography dependency on rust | 17:32 |
| mgoddard | fix for a similar issue in tenks: https://review.opendev.org/c/openstack/tenks/+/774474 | 17:32 |
| mgoddard | (in case anyone is looking at it) | 17:32 |
| priteau | I wish there was a newer pip we could install as RPM on RHEL | 17:41 |
| *** k_mouza_ has joined #openstack-kolla | 17:42 | |
| *** k_mouza_ has quit IRC | 17:42 | |
| *** k_mouza has quit IRC | 17:44 | |
| *** maharg101 has quit IRC | 18:01 | |
| *** gfidente is now known as gfidente|afk | 18:06 | |
| *** dougsz has quit IRC | 18:11 | |
| *** SmearedBeard has joined #openstack-kolla | 18:17 | |
| *** emccormick has quit IRC | 18:25 | |
| *** emccormick has joined #openstack-kolla | 18:26 | |
| *** bengates_ has joined #openstack-kolla | 18:29 | |
| *** bengates has quit IRC | 18:33 | |
| *** hjensas|afk is now known as hjensas | 18:37 | |
| SmearedBeard | Hi all.. I'm just reading about Kayobe, and wasn't sure about the supported OS mentioned in https://docs.openstack.org/kayobe/latest/support-matrix.html. is this just the OS that Keyboe is installed to and runs from? | 19:08 |
| openstackgerrit | Dincer Celik proposed openstack/kolla-ansible stable/victoria: Fix Barbican API log config https://review.opendev.org/c/openstack/kolla-ansible/+/774484 | 19:12 |
| openstackgerrit | Dincer Celik proposed openstack/kolla-ansible stable/ussuri: Fix Barbican API log config https://review.opendev.org/c/openstack/kolla-ansible/+/774485 | 19:14 |
| openstackgerrit | Giacomo Lanciano proposed openstack/kolla-ansible master: Increment monasca-grafana check retries https://review.opendev.org/c/openstack/kolla-ansible/+/774531 | 19:15 |
| yoctozepto | 18:32:23 <mgoddard> kolla-ansible CI broken by python cryptography dependency on rust | 19:21 |
| yoctozepto | and python 4 has not started its reign yet! :D | 19:21 |
| *** e0ne has joined #openstack-kolla | 19:28 | |
| *** e0ne has quit IRC | 19:28 | |
| *** e0ne has joined #openstack-kolla | 19:38 | |
| *** k_mouza has joined #openstack-kolla | 19:44 | |
| *** k_mouza has quit IRC | 19:48 | |
| *** SmearedBeard has quit IRC | 19:56 | |
| *** maharg101 has joined #openstack-kolla | 19:58 | |
| *** maharg101 has quit IRC | 20:02 | |
| openstackgerrit | Radosław Piliszek proposed openstack/kolla-ansible stable/train: Fix Barbican API log config https://review.opendev.org/c/openstack/kolla-ansible/+/774486 | 20:06 |
| openstackgerrit | Radosław Piliszek proposed openstack/kolla-ansible stable/train: Fix Barbican API log config https://review.opendev.org/c/openstack/kolla-ansible/+/774486 | 20:11 |
| *** cah_link has joined #openstack-kolla | 20:15 | |
| *** kevko has quit IRC | 20:25 | |
| *** gfidente|afk has quit IRC | 20:32 | |
| *** e0ne has quit IRC | 20:34 | |
| openstackgerrit | Merged openstack/kayobe stable/train: Fix --limit with commas https://review.opendev.org/c/openstack/kayobe/+/770960 | 20:38 |
| *** k_mouza has joined #openstack-kolla | 20:59 | |
| *** k_mouza has quit IRC | 21:04 | |
| *** jonaspaulo has quit IRC | 21:09 | |
| *** cah_link has quit IRC | 21:10 | |
| *** jbadiapa has quit IRC | 21:11 | |
| *** bengates_ has quit IRC | 22:10 | |
| *** bengates has joined #openstack-kolla | 22:46 | |
| *** parallax has quit IRC | 23:04 | |
| *** snbuback has quit IRC | 23:04 | |
| *** snbuback has joined #openstack-kolla | 23:05 | |
| *** parallax has joined #openstack-kolla | 23:05 | |
| *** ebbex has quit IRC | 23:14 | |
| *** ebbex has joined #openstack-kolla | 23:36 | |
| *** bengates has quit IRC | 23:38 | |
| openstackgerrit | Merged openstack/kayobe stable/train: Fix /tmp/swift-rings/backups/ files deletion https://review.opendev.org/c/openstack/kayobe/+/771928 | 23:51 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!