| *** threestrands has joined #openstack-kolla | 00:35 | |
| *** k_mouza has joined #openstack-kolla | 02:17 | |
| *** xinliang has joined #openstack-kolla | 02:18 | |
| *** k_mouza has quit IRC | 02:22 | |
| *** k_mouza has joined #openstack-kolla | 02:26 | |
| *** k_mouza has quit IRC | 02:33 | |
| *** k_mouza has joined #openstack-kolla | 02:38 | |
| openstackgerrit | ya.wang proposed openstack/kolla-ansible master: Fix that cyborg agent failed to start privesp daemon. https://review.opendev.org/721139 | 02:38 |
|---|---|---|
| *** k_mouza has quit IRC | 02:46 | |
| *** k_mouza has joined #openstack-kolla | 02:52 | |
| *** k_mouza has quit IRC | 02:57 | |
| *** k_mouza has joined #openstack-kolla | 03:01 | |
| *** kevinz has joined #openstack-kolla | 03:06 | |
| *** k_mouza has quit IRC | 03:08 | |
| *** k_mouza has joined #openstack-kolla | 03:15 | |
| openstackgerrit | ya.wang proposed openstack/kolla-ansible master: Fix taht cyborg conductor failed to communicate with placement https://review.opendev.org/721158 | 03:15 |
| *** k_mouza has quit IRC | 03:20 | |
| *** k_mouza has joined #openstack-kolla | 03:22 | |
| *** k_mouza has quit IRC | 03:28 | |
| *** k_mouza has joined #openstack-kolla | 03:31 | |
| *** k_mouza has quit IRC | 03:37 | |
| *** k_mouza has joined #openstack-kolla | 03:40 | |
| *** k_mouza has quit IRC | 03:45 | |
| *** ricolin has joined #openstack-kolla | 03:48 | |
| *** k_mouza has joined #openstack-kolla | 04:02 | |
| *** ykarel|away is now known as ykarel | 04:09 | |
| *** k_mouza has quit IRC | 04:10 | |
| *** xinliang has quit IRC | 04:11 | |
| *** k_mouza has joined #openstack-kolla | 04:16 | |
| *** k_mouza has quit IRC | 04:20 | |
| *** k_mouza has joined #openstack-kolla | 04:22 | |
| *** evrardjp has quit IRC | 04:23 | |
| *** k_mouza has quit IRC | 04:27 | |
| *** skramaja has joined #openstack-kolla | 04:29 | |
| *** k_mouza has joined #openstack-kolla | 04:38 | |
| *** k_mouza has quit IRC | 04:43 | |
| *** k_mouza has joined #openstack-kolla | 04:46 | |
| *** k_mouza has quit IRC | 04:50 | |
| *** k_mouza has joined #openstack-kolla | 04:56 | |
| *** cah_link has joined #openstack-kolla | 05:00 | |
| *** k_mouza has quit IRC | 05:02 | |
| *** k_mouza has joined #openstack-kolla | 05:13 | |
| *** k_mouza has quit IRC | 05:17 | |
| *** k_mouza has joined #openstack-kolla | 05:23 | |
| *** k_mouza has quit IRC | 05:28 | |
| *** k_mouza has joined #openstack-kolla | 05:33 | |
| *** k_mouza has quit IRC | 05:38 | |
| *** prernadembla05 has joined #openstack-kolla | 05:39 | |
| *** k_mouza has joined #openstack-kolla | 05:41 | |
| *** evrardjp has joined #openstack-kolla | 05:41 | |
| *** k_mouza has quit IRC | 05:45 | |
| *** vishalmanchanda has joined #openstack-kolla | 05:46 | |
| *** shyamb has joined #openstack-kolla | 05:48 | |
| osmanlicilegi | morning | 05:49 |
| *** k_mouza has joined #openstack-kolla | 05:51 | |
| *** k_mouza has quit IRC | 05:55 | |
| *** shyamb has quit IRC | 06:04 | |
| *** seco has joined #openstack-kolla | 06:04 | |
| *** shyamb has joined #openstack-kolla | 06:05 | |
| *** k_mouza has joined #openstack-kolla | 06:11 | |
| *** k_mouza has quit IRC | 06:16 | |
| *** numans has joined #openstack-kolla | 06:29 | |
| *** jbadiapa has joined #openstack-kolla | 06:49 | |
| yoctozepto | morning | 06:51 |
| *** shyamb has quit IRC | 06:53 | |
| *** shyamb has joined #openstack-kolla | 06:53 | |
| *** k_mouza has joined #openstack-kolla | 06:56 | |
| *** k_mouza has quit IRC | 07:01 | |
| *** shyamb has quit IRC | 07:03 | |
| *** dougsz has joined #openstack-kolla | 07:05 | |
| openstackgerrit | Radosław Piliszek proposed openstack/kolla-ansible master: Zun: Add zun-cni-daemon to compute node https://review.opendev.org/708213 | 07:10 |
| *** bengates has joined #openstack-kolla | 07:11 | |
| *** rpittau|afk is now known as rpittau | 07:12 | |
| *** hjensas has quit IRC | 07:13 | |
| *** hjensas has joined #openstack-kolla | 07:16 | |
| *** threestrands_ has joined #openstack-kolla | 07:18 | |
| *** threestrands_ has quit IRC | 07:18 | |
| *** bengates has quit IRC | 07:21 | |
| *** threestrands has quit IRC | 07:21 | |
| *** bengates has joined #openstack-kolla | 07:21 | |
| sri_ | morning | 07:21 |
| *** k_mouza has joined #openstack-kolla | 07:31 | |
| *** spsurya_ has joined #openstack-kolla | 07:35 | |
| *** k_mouza has quit IRC | 07:36 | |
| mnasiadka | morning | 07:46 |
| *** negronjl has quit IRC | 07:58 | |
| *** xinliang has joined #openstack-kolla | 08:00 | |
| openstackgerrit | Michal Nasiadka proposed openstack/kolla-ansible master: WIP: Ansible lint related fixes https://review.opendev.org/721191 | 08:01 |
| *** negronjl has joined #openstack-kolla | 08:01 | |
| *** xinliang has quit IRC | 08:12 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Manage nova scheduler workers count https://review.opendev.org/721198 | 08:13 |
| *** ykarel is now known as ykarel|lunch | 08:17 | |
| openstackgerrit | Michal Nasiadka proposed openstack/kolla-ansible master: CI: Add ansible-lint to tox https://review.opendev.org/694779 | 08:17 |
| *** shyamb has joined #openstack-kolla | 08:23 | |
| mgoddard | morning | 08:29 |
| sri_ | mnasiadka: hi , quick question, Is ceph-ansible supported in ussuri release ? | 08:42 |
| sri_ | mnasiadka: from kolla-ansible | 08:42 |
| mnasiadka | sri_: external ceph is supported in Ussuri release (no more kolla ceph deployment), you can use any deployment tool - we use ceph-ansible in the CI for testing | 08:43 |
| *** shyam89 has joined #openstack-kolla | 08:45 | |
| sri_ | mnasiadka: ack, is there any way to tell ceph-ansible to skip the docker install ? I've tried to " --skip-tags docker" but it's not working, | 08:47 |
| mnasiadka | sri_: --skip-tags=with_pkg | 08:47 |
| *** shyamb has quit IRC | 08:48 | |
| sri_ | mnasiadka: Thanks :) | 08:50 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible master: Manage nova scheduler workers count https://review.opendev.org/721198 | 08:51 |
| *** vishalmanchanda has quit IRC | 08:56 | |
| *** ykarel|lunch is now known as ykarel | 09:00 | |
| *** k_mouza has joined #openstack-kolla | 09:00 | |
| mgoddard | thanks for the ML post yoctozepto. I'm going to let it sink for a while before replying | 09:03 |
| hrw | elo | 09:05 |
| oklhost | Hi @all. Anybody facing issues with latest Google Chrome and "Launch Instance" button in horizon? The modal won't open... | 09:06 |
| hrw | oklhost: #openstack-horizon? | 09:08 |
| oklhost | thx | 09:08 |
| *** shyam89 has quit IRC | 09:20 | |
| *** shyam89 has joined #openstack-kolla | 09:20 | |
| openstackgerrit | Radosław Piliszek proposed openstack/kolla-ansible master: Zun: Add zun-cni-daemon to compute node https://review.opendev.org/708213 | 09:25 |
| yoctozepto | mgoddard: thanks, I had your poc starred already :-) | 09:27 |
| *** ricolin has quit IRC | 09:28 | |
| openstackgerrit | Marcin Juszkiewicz proposed openstack/kolla-ansible master: CI: add Debian source upgrade job on AArch64 https://review.opendev.org/713854 | 09:30 |
| openstackgerrit | Marcin Juszkiewicz proposed openstack/kolla-ansible master: CI: add Debian source upgrade job on AArch64 https://review.opendev.org/713854 | 09:30 |
| hrw | rebased and dropped depends-on | 09:30 |
| *** shyam89 has quit IRC | 09:32 | |
| *** shyamb has joined #openstack-kolla | 09:35 | |
| hrw | yoctozepto: +2 on both python27 cleanups | 09:37 |
| *** e0ne has joined #openstack-kolla | 09:42 | |
| *** shyam89 has joined #openstack-kolla | 09:45 | |
| *** shyamb has quit IRC | 09:48 | |
| *** ricolin has joined #openstack-kolla | 09:48 | |
| *** ricolin_ has joined #openstack-kolla | 09:48 | |
| yoctozepto | hrw: thanks, k-a has to wait for fixing (and need to check why infra c8 images are far from vanilla - I love how introducing bugs reveals other bugs too) | 09:53 |
| *** vishalmanchanda has joined #openstack-kolla | 09:53 | |
| *** gfidente has joined #openstack-kolla | 09:55 | |
| stingrayza | hey all | 09:57 |
| stingrayza | my k-a isn't picking up my nova.conf override in /etc/kolla/config/nova/nova.conf | 09:58 |
| stingrayza | I'm doing a "kolla-ansible reconfigure --tags nova" - that should pick up the file and run with it, right? | 09:59 |
| yoctozepto | nova.conf or nova/nova-api.conf etc. | 10:00 |
| yoctozepto | many folks seem to be mistaken, maybe we should just pick up either... | 10:01 |
| *** ricolin_ has quit IRC | 10:02 | |
| stingrayza | https://docs.openstack.org/kolla-ansible/latest/admin/advanced-configuration.html#openstack-service-configuration-in-kolla might need an update - that's perhaps the source of the confusion...? | 10:02 |
| * stingrayza tries again with the file one level higher... | 10:04 | |
| *** e0ne has quit IRC | 10:05 | |
| *** e0ne has joined #openstack-kolla | 10:05 | |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Cleanup py27 support https://review.opendev.org/720764 | 10:09 |
| *** e0ne has quit IRC | 10:12 | |
| *** e0ne has joined #openstack-kolla | 10:13 | |
| *** rpittau is now known as rpittau|bbl | 10:15 | |
| openstackgerrit | ya.wang proposed openstack/kolla-ansible master: Fix that cyborg conductor failed to communicate with placement https://review.opendev.org/721158 | 10:25 |
| *** shyam89 has quit IRC | 10:33 | |
| openstackgerrit | ya.wang proposed openstack/kolla-ansible master: Fix that cyborg agent failed to start privsep daemon. https://review.opendev.org/721139 | 10:48 |
| stingrayza | thanks yoctozepto - moving it 'up' one dir worked. I'm gonna log a bug for a fix (and take it) | 10:50 |
| stingrayza | logged (https://bugs.launchpad.net/kolla-ansible/+bug/1873782) | 10:51 |
| openstack | Launchpad bug 1873782 in kolla-ansible "nova.conf override not used from documented location" [Undecided,New] - Assigned to Raimund Hook (stingrayza) | 10:51 |
| *** e0ne has quit IRC | 10:51 | |
| *** e0ne has joined #openstack-kolla | 10:52 | |
| stingrayza | happy to fix this, - the question is do I just do a quick doc clean-up, or do we fix the conf pick up to additionally pick up "{{ node_custom_config }}/nova/nova.conf" ? (and if the latter, then do I do that for _all_ services?) | 10:54 |
| *** dougsz has quit IRC | 11:02 | |
| mgoddard | stingrayza: I think typically we don't support that path, although there may be exceptions. Worth checking a few to be sure, but I'd update docs | 11:08 |
| stingrayza | mgoddard: cool. doc update inbound :) | 11:10 |
| *** dougsz has joined #openstack-kolla | 11:17 | |
| hrw | meh. | 11:18 |
| hrw | I look at kolla-toolbox and while it can be moved to use binary packages for most of stuff the problem stays due to ubuntu shipping ancient ansible ;( | 11:19 |
| hrw | debian:buster has 2.9.6 in backports, centos has 2.9.7 in some repos | 11:20 |
| hrw | and it would also allow to drop compiler from it | 11:21 |
| hrw | or we add https://launchpad.net/~ansible/+archive/ubuntu/ansible and use binary package 2.9.x | 11:23 |
| openstackgerrit | Raimund Hook (stingrayza) proposed openstack/kolla-ansible master: Update Advanced Config guide to clarify paths https://review.opendev.org/721241 | 11:25 |
| mgoddard | hrw: true, but ansible doesn't require upper constraints | 11:27 |
| *** hjensas has quit IRC | 11:28 | |
| *** sean-k-mooney has joined #openstack-kolla | 11:29 | |
| hrw | mgoddard: and we want 2.8+ while bionic has 2.5 | 11:31 |
| mgoddard | hrw: yeah, I'm saying we could just pip install ansible | 11:31 |
| mgoddard | hrw: will still be an improvement over today | 11:32 |
| hrw | mgoddard: and then we have to deal with virtualenv paths so have to install all python stuff in venv. which means building them. installing compilers, headers... | 11:32 |
| mnasiadka | hrw: well, using the ppa is not bad I guess, especially that it seems Red Hat is maintaining it? | 11:32 |
| mgoddard | hrw: we wouldn't use a virtualenv for binary, install ansible to system | 11:32 |
| hrw | mgoddard: kolla-toolbox is infra image. | 11:33 |
| mgoddard | hrw: not yet :) | 11:33 |
| hrw | it is. just not called that | 11:33 |
| mgoddard | but I see your point | 11:33 |
| hrw | normal/debian-binary-kolla-toolbox 9.1.0 0d7a331a881b 29 minutes ago 979MB | 11:37 |
| hrw | binaryonly/debian-binary-kolla-toolbox 9.1.0 637522ed9bf9 5 seconds ago 619MB | 11:37 |
| hrw | 360MB | 11:37 |
| *** e0ne has quit IRC | 11:43 | |
| *** e0ne has joined #openstack-kolla | 11:43 | |
| hrw | INFO:kolla.common.utils.kolla-toolbox:Setting up ansible (2.9.7-1ppa~bionic) ... | 11:43 |
| hrw | binaryonly/ubuntu-binary-kolla-toolbox 9.1.0 d73aed9b0b32 29 seconds ago 648MB | 11:44 |
| *** dking has joined #openstack-kolla | 11:48 | |
| *** Fl1nt has joined #openstack-kolla | 11:50 | |
| Fl1nt | Hi everyone! | 11:51 |
| Fl1nt | How are you? | 11:51 |
| hrw | mgoddard: hm. what for we needed python-mysqlclient in k-a? | 11:52 |
| dking | Good morning/evening, Fl1nt. I'm just getting started for the day early. | 11:52 |
| hrw | mgoddard: it is the only thing missing in centos ;d | 11:52 |
| mgoddard | hrw: pass | 11:52 |
| mgoddard | hrw: https://docs.ansible.com/ansible/latest/modules/mysql_db_module.html | 11:53 |
| mgoddard | uses PyMySQL | 11:53 |
| Fl1nt | @dking, it's the afternoon in here ^^ | 11:53 |
| hrw | gracias | 11:53 |
| dking | Fl1nt: Then, that might make it easier for in this channel. I find that folks are usually going away about the time that I get started for the day. | 11:55 |
| hrw | ~curse rpm distros for PacKAgeSwIThuPPerCAseLETTersInNamES | 11:56 |
| hrw | Fl1nt, dking: google for UGT. | 11:57 |
| hrw | Universal Greeting Time | 11:57 |
| dking | Kayobe question: So, I'm a bit more familiar with the configuration now. However, when I mentioned before that I was wanting multiple isolated overclouds, I may have forgotten to mention that I was wanting those overclouds to be able to be unique. It looks like the configuration here implies that they are all the same kolla-ansible setup? Would that mean that I would have to deploy them as baremetal servers instead? | 11:57 |
| Fl1nt | @dking, that's for sure as many of the active members are actually located un europe. | 11:58 |
| hrw | binaryonly/centos-binary-kolla-toolbox 9.1.0 321b14f10ad7 48 seconds ago 769MB | 11:58 |
| *** e0ne has quit IRC | 11:58 | |
| dking | hrw: That's pretty useful. I suppose that it's a different world these days. I think I'll adopt that. | 11:58 |
| Fl1nt | @dking, you need to setup one repository per overcloud that you want to deploy | 11:59 |
| hrw | dking: it simplifies greetings a lot | 11:59 |
| mgoddard | dking: Fl1nt is right - use a separate kayobe-config repo/branch per environment | 12:00 |
| dking | Is that one kayobe-config directory per overcloud? | 12:00 |
| mgoddard | dking: yes | 12:00 |
| dking | Okay. That might work, then. | 12:00 |
| mgoddard | dking: we have a feature proposed to support multiple envs in one, but haven't got there yet | 12:00 |
| Fl1nt | that's simplier, cleaner and except if you're Google's sized, you shouldn't have hundreds of overclouds ^^ | 12:00 |
| hrw | mgoddard: now kolla-toolbox has few layers less too due to all RUN calls done at once | 12:01 |
| dking | Another question, then. I see that it mentions how to setup a bond, and then how to setup VLAN. Each of my servers have two NICs that need to be bonded, and then will use different VLANs on that bond. Would I setup a dummy interface to make the bond, and then set the VLANs for the actual *_interface? | 12:02 |
| mgoddard | hrw: nice | 12:02 |
| dking | Fl1nt: Actually, even though we're no where near Google sized, we do plan to have hundreds of overclouds. We're looking at doing many smaller isolated overclouds. | 12:03 |
| *** rpittau|bbl is now known as rpittau | 12:03 | |
| mgoddard | dking: it depends if you want an IP on the native VLAN. If so, that network will define the bond. Otherwise you'll need a dummy to represent the bond | 12:04 |
| hrw | mgoddard: 5 files changed, 35 insertions(+), 64 deletions(-) | 12:04 |
| mgoddard | dking: another thing to consider - for OVS we currently require a bridge, and kayobe will plumb a veth pair from that bridge to the OVS bridge | 12:04 |
| dking | mgoddard: Do you have some info on that project? I won't need it starting out, but if we do end up going with Kayobe, that's something we would likely be interested in. | 12:05 |
| mgoddard | dking: this patch allows you to plug an interface straight into OVS: https://review.opendev.org/705037 | 12:05 |
| patchbot | patch 705037 - kayobe - Allow OVS bridges to connect directly to interface - 6 patch sets | 12:05 |
| Fl1nt | @dking, ok depending on you constraints regarding isolation/security I do understand that plan, however, I can advise you from experience, that using that much spread isn't going to be easy to manage for operators, you would likely rather have a giant overcloud with segmentation | 12:05 |
| Fl1nt | using | 12:05 |
| Fl1nt | cell/zone/region/aggregates | 12:05 |
| mgoddard | dking: here's the story for multiple envs: https://storyboard.openstack.org/#!/story/2002009 | 12:06 |
| dking | That's great! I'm sure that we'll be looking at that a little more soon. | 12:10 |
| dking | Fl1nt: Yes, that's true. But we're actually wanting to sell overclouds. And yes, that is going to be a lot to manage, but that's part of the new direction. | 12:12 |
| Fl1nt | @dking, oh ok, I do understand your business case, make sense and that's actually easier to manager with one/x overcloud splitted per customer as you can use your repository project/subproject tree to cleanly store them. | 12:14 |
| dking | mgoddard: I'm having a bit of trouble understanding the part about OVS. Once we get the overclouds up, we won't want them using the Ansible or seed hosts for the networking, so would that still apply? Of course, the overcloud controllers will eventually be setting up their own OVS for Neutron. | 12:14 |
| mgoddard | dking: I'm referring to the OVS networking within a cloud. No seed involved. It's just about how the network interfaces need to be wired up | 12:16 |
| dking | Fl1nt: Yes, that's one way. We've actually been tossing around ideas about different deploy methods, and there's even some consideration given to using VMs for deploy. There will be an entire separate system for managing the details, regardless how we implement the deploy. | 12:16 |
| mgoddard | dking: with a bond, you'll want physical -> bond -> bridge | 12:16 |
| mgoddard | then kayobe will plumb a veth from that bridge to OVS | 12:16 |
| openstackgerrit | Marcin Juszkiewicz proposed openstack/kolla master: Debian/Ubuntu: use Ansible 2.9+ binary packages https://review.opendev.org/721251 | 12:17 |
| openstackgerrit | Marcin Juszkiewicz proposed openstack/kolla master: kolla-toolbox: use only binary packages https://review.opendev.org/721252 | 12:17 |
| hrw | let CI test was I right ;d | 12:17 |
| Fl1nt | @dking, interesting ^^ let us know about how things evolves ^^ | 12:18 |
| dking | mgoddard: At the moment, the physical networks are connected at the switch using port groups. | 12:18 |
| mgoddard | dking: when I say bridge, I mean a linux bridge not hardware | 12:19 |
| dking | Up to this point, I don't think that I've been using a bridge, other than the bridges that Neutron sets up. This is a way that Kayobe handles networking while deploying, and this patch allows us to bypass that because we don't need an IP on the native VLAN? | 12:27 |
| Fl1nt | @dking, even a bond don't need an IP, only your vnic need one. | 12:30 |
| dking | I'm going to be needing IPs on the various VLAN interfaces on the bond, corresponding to the various networks. I just won't need an IP on the native VLAN. | 12:32 |
| yoctozepto | mgoddard, hrw: I see kolla-toolbox relying on ansible as a kind of accidental complexity; I mean, what kolla-toolbox does is perfectly doable without the weaponry of ansible | 12:32 |
| Fl1nt | yeah that's ok, just set up your vlan interfaces. | 12:32 |
| hrw | yoctozepto: ansible/roles/common/handlers/main.yml: command: docker exec -t "{{ common_services['kolla-toolbox']['container_name'] }}" ansible --version | 12:35 |
| yoctozepto | stingrayza: for the love of kolla, I did not notice there was this outrageous mistake in the docs, yikes | 12:35 |
| hrw | yoctozepto: so it is used :D | 12:35 |
| yoctozepto | hrw: yeah, it is used | 12:35 |
| yoctozepto | hrw: I meant to say it does not have to be by design | 12:35 |
| hrw | yoctozepto: you know that I try to stay away from k-a as much as possible ;D | 12:35 |
| yoctozepto | hrw: don't drop ansible right away :D | 12:35 |
| yoctozepto | hrw: sure, I know | 12:36 |
| hrw | yoctozepto: I do not plan to drop it D: | 12:36 |
| dking | Okay. Another question. I've got the seed deployed and running. I'm going to be trying to set up a single control node in the overcloud to test things out. Because we're still using the switch for some other things at the moment, I'm not going to be using discover, but manually adding the server with kolla_bifrost_servers. Would I still need to make entries for the physical network? At the moment, my next step is real | 12:37 |
| dking | see the node available in bifrost. | 12:37 |
| dking | (or technically, even manageable, for the moment) | 12:38 |
| *** Luzi has joined #openstack-kolla | 12:42 | |
| *** priteau has joined #openstack-kolla | 12:50 | |
| stingrayza | yoctozepto: hopefully that will fix the 'many folks seem to be mistaken' :) | 12:55 |
| *** ykarel is now known as ykarel|afk | 12:56 | |
| openstackgerrit | Marcin Juszkiewicz proposed openstack/kolla-ansible master: Improve CI ansible rules to pass ansible-lint https://review.opendev.org/720664 | 13:02 |
| hrw | improved commit message a bit | 13:02 |
| *** irclogbot_0 has quit IRC | 13:06 | |
| *** irclogbot_1 has joined #openstack-kolla | 13:07 | |
| yoctozepto | stingrayza: hahaha, yeah | 13:08 |
| *** kevinz has quit IRC | 13:09 | |
| *** kevinz has joined #openstack-kolla | 13:14 | |
| hrw | Switched to a new branch 'wip/train-use-c-r-o-t' | 13:16 |
| hrw | now just have to wait for c-r-o-t package to appear ;d | 13:16 |
| *** skramaja has quit IRC | 13:17 | |
| *** dswebb has joined #openstack-kolla | 13:22 | |
| dswebb | Hi all, quick one. just saw that OVN support has been merged into master, is there any plan to back port this to train or will it be a ussari only feature? | 13:23 |
| kplant | usually features aren't backported, only fixes; i'll let someone else confirm | 13:24 |
| yoctozepto | dswebb: ussuri only | 13:28 |
| yoctozepto | (and forward of course) | 13:28 |
| dswebb | perfect, cheers | 13:42 |
| *** TrevorV has joined #openstack-kolla | 13:50 | |
| *** prernadembla05 has quit IRC | 13:52 | |
| mgoddard | dking: to be honest I haven't used kolla_bifrost_servers in a long time. You could give it a try, or if you are familiar with ironic just enroll the node manually | 13:52 |
| mgoddard | dking: what do you mean by 'make entries for the phyiscal network'? As in switches? | 13:53 |
| -openstackstatus- NOTICE: Zuul is temporarily offline; service should be restored in about 15 minutes. | 13:59 | |
| *** hongbin has joined #openstack-kolla | 14:21 | |
| *** ykarel|afk is now known as ykarel | 14:22 | |
| *** cah_link has quit IRC | 14:27 | |
| *** cdearborn has joined #openstack-kolla | 14:34 | |
| dmsimard | yoctozepto: btw, reversed the truncate for improved relevantness https://api.trunk.demo.recordsansible.org/ | 14:42 |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Fix passwords.yml generation with vault encryption on Python 3 https://review.opendev.org/721281 | 14:43 |
| yoctozepto | dmsimard: yay, looks way better, thanks | 14:44 |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Cleanup py27 support https://review.opendev.org/720764 | 14:47 |
| *** jimcrowleyibm has joined #openstack-kolla | 14:49 | |
| Fl1nt | @mgoddard, so, I've just stumble upon a funny side effect of k-a that plan to not deploy/manage ceph anymore in futur release. | 14:50 |
| Fl1nt | as our infrastructure is "hyperconverged" we've all our openstack host to install ceph using ceph-ansible with a containerized deployment. | 14:51 |
| Fl1nt | C-A use python-docker-py package from YUM (installed system wide) and K-A use docker pip package. | 14:52 |
| Fl1nt | both packages conflict | 14:52 |
| Fl1nt | as python-docker-py take precedence over docker pip package | 14:53 |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Revove support for CentOS 7 and Python 2 https://review.opendev.org/720818 | 14:53 |
| Fl1nt | when performing a precheck step your deployment fail as docker.__version__ is 1.10.6 instead of > 2.x | 14:53 |
| *** hjensas has joined #openstack-kolla | 14:54 | |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Remove support for CentOS 7 and Python 2 https://review.opendev.org/720818 | 14:54 |
| mgoddard | Fl1nt: :( haven't come across that one before | 14:55 |
| Fl1nt | sooo, all in all, I'll fill an issue with C-A as they don't use venv and just rely on YUM by assuming they're the only citizen in town, which isn't really gentle at all ^^ | 14:55 |
| mgoddard | python-docker-py is pretty old, sure there is no way to configure it? | 14:55 |
| Fl1nt | unfortunately not, it's hardcoded within the prerequisite role | 14:55 |
| yoctozepto | Fl1nt: k-a can use virtualenv | 14:56 |
| mgoddard | Fl1nt: you can use a virtualenv for k-a | 14:56 |
| Fl1nt | it need a patch from C-A tbh | 14:56 |
| Fl1nt | I do | 14:56 |
| yoctozepto | mgoddard: agreed then | 14:56 |
| mgoddard | Fl1nt: for remote hosts? | 14:56 |
| yoctozepto | Fl1nt: on the target hosts | 14:56 |
| yoctozepto | pff | 14:56 |
| yoctozepto | :D | 14:56 |
| mgoddard | yoctozepto: too slow | 14:56 |
| Fl1nt | hehehe let me check. | 14:56 |
| mgoddard | I assume yoctozepto is an echo bot that just rephrases what I say | 14:56 |
| yoctozepto | mgoddard: nah, I was second faster on the first statement | 14:57 |
| mgoddard | hmm, right | 14:57 |
| mgoddard | time travelling echo bot? | 14:57 |
| yoctozepto | count me in! | 14:57 |
| Fl1nt | hum... ok, seems we don't use venv, which vars define that ? | 14:58 |
| Fl1nt | that could do the trick | 14:58 |
| Fl1nt | but yet | 14:58 |
| Fl1nt | C-A shouldn't assume to be the only one to live in host especially when dealing with containerized installation. | 14:58 |
| yoctozepto | Fl1nt: it shouldn't | 14:58 |
| yoctozepto | Fl1nt: but it's not our baby :-) | 14:59 |
| Fl1nt | yep | 14:59 |
| sri_ | Fl1nt: I've faced that issue when i am deploying c-a and k-a in the same nodes | 14:59 |
| yoctozepto | you can go there and tell folks kolla boys are doing it right ;D | 14:59 |
| Fl1nt | ok, I'll be back in few hours, happy to propose something regarding ceph as containers images are pretty solid. | 14:59 |
| Fl1nt | @sri_, yep, that's the scenario. | 15:00 |
| sri_ | the quick fix found is pip uninstall all docker packages, exp: docker, docker-py, docker-*, then run the pip install docker | 15:01 |
| mgoddard | Fl1nt: https://docs.openstack.org/kolla-ansible/latest/user/virtual-environments.html | 15:01 |
| mgoddard | Fl1nt: kayobe does this by default, which might explain why I haven't seen it | 15:01 |
| kplant | ^ that's how i run my clouds | 15:04 |
| kplant | much safer imo | 15:04 |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Remove nameservers with any IP in overcloud resolv.conf workaround https://review.opendev.org/721285 | 15:06 |
| sri_ | mgoddard: this looks more safer and cleaner, thanks for link :) | 15:07 |
| dmsimard | mgoddard: btw during the weekend I sent https://review.opendev.org/#/c/720902/ to test the newest release of ara, this is the data we've got across all jobs: https://api.trunk.demo.recordsansible.org/?name=&path=kolla-ansible&label= | 15:07 |
| patchbot | patch 720902 - kolla-ansible - DNM: Test ara 1.4 with kolla-ansible - 1 patch set | 15:07 |
| dmsimard | the equivalent in API form: https://api.trunk.demo.recordsansible.org/api/v1/playbooks?path=kolla-ansible | 15:07 |
| *** born2bake has joined #openstack-kolla | 15:07 | |
| *** dougsz has quit IRC | 15:09 | |
| mgoddard | dmsimard: neat, looks nice! | 15:09 |
| dking | mgoddard: Sorry, I was away. Yes, I meant switches. | 15:18 |
| *** ykarel is now known as ykarel|away | 15:18 | |
| mgoddard | dking: ok, that's all optional - kayobe doesn't assume it manages them. You can still run discovery | 15:18 |
| mgoddard | dking: the requirements are the same as for provisioning - PXE boot | 15:19 |
| mgoddard | dking: assume your bonds have some fallback config for PXE? | 15:19 |
| dking | No, PXE only understands unbonded, untagged. I was assuming that Kayobe would be creating the bonds/VLANs after booting from PXE. | 15:20 |
| *** well100 has quit IRC | 15:21 | |
| mgoddard | dking: I mean on the switch side | 15:22 |
| mgoddard | hrw, yoctozepto, mnasiadka, osmanlicilegi: let's land this thing: https://review.opendev.org/#/c/713283/ https://review.opendev.org/#/c/713282/ | 15:26 |
| patchbot | patch 713283 - kolla-ansible - Remove support for Python 2 on the host - 5 patch sets | 15:26 |
| patchbot | patch 713282 - kolla-ansible - Remove support for CentOS 7 - 2 patch sets | 15:26 |
| dking | Yes, they should be able to handle the packets without a bond for PXE. We have a separate switch for the BMC network and ports. The servers can also receive untagged packets to a single interface. | 15:26 |
| *** vishalmanchanda has quit IRC | 15:26 | |
| mgoddard | dking: ok, should be fine | 15:26 |
| *** hongbin has quit IRC | 15:28 | |
| hrw | mgoddard: next step: remove easy_install | 15:31 |
| mgoddard | hrw: go for it :) | 15:31 |
| hrw | +2 on both | 15:32 |
| cosmicsound | dmsimard , will ara be included in kolla-ansible? | 15:37 |
| mgoddard | hrw: thanks | 15:37 |
| dmsimard | cosmicsound: define included ? it's enabled in CI but other than that I am not aware | 15:38 |
| dmsimard | like there could be a kolla container with the ara API server in it maybe ? | 15:38 |
| cosmicsound | yep, thats what i meant | 15:38 |
| openstackgerrit | Marcin Juszkiewicz proposed openstack/kolla-ansible master: Drop handling of 'easy_install' https://review.opendev.org/721290 | 15:40 |
| dmsimard | cosmicsound: it's not on my personal to-do list at this time but if anyone wants to do it I can point them in the right direction | 15:40 |
| dmsimard | fwiw I am testing container image releases, this is the first one: https://hub.docker.com/r/recordsansible/ara-api | 15:40 |
| dmsimard | built from https://review.opendev.org/#/c/687905/5/contrib/containers/ara-api-container-source.sh | 15:41 |
| patchbot | patch 687905 - recordsansible/ara - Add scripts to build and run ARA API containers - 5 patch sets | 15:41 |
| dking | Great. I'm giving it a try. I'm receiving an error for the overcloud node deploy because kolla_external_vip_interface is not set. kolla_external_vip_interface is only set when {{ public_net_name }} is in network_interfaces. network_interfaces seems like it should include public_net_name. I'm trying to track things back, but if anybody has a hint at what I might have left out, I would appreciate it. | 15:42 |
| *** happyhemant has joined #openstack-kolla | 15:45 | |
| -openstackstatus- NOTICE: Gerrit will be restarted to correct a misconfiguration which caused some git mirrors to have outdated references. | 15:47 | |
| mgoddard | dking: public_net_name should be in network_interfaces if the host is in the 'network' group | 15:48 |
| dking | Yes, I'm seeing that. I'm trying to dump variables. I see that groups['network'] contains the host. | 15:53 |
| dking | Odd. controller_network_host_network_interfaces is empty. | 15:53 |
| *** hjensas has quit IRC | 15:54 | |
| openstackgerrit | Merged openstack/kolla master: Cleanup py27 support https://review.opendev.org/720751 | 15:55 |
| dking | So, for some reason, this: "{{ ([public_net_name, tunnel_net_name] + external_net_names) | reject('none') | unique | list }}" is empty, even though public_net_name is "external". I'm dumping that just a few lines after I print out public_net_name. | 15:56 |
| *** cah_link has joined #openstack-kolla | 15:56 | |
| *** KeithMnemonic has joined #openstack-kolla | 16:00 | |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: CentOS 8: separate kolla build tag from deploy tag https://review.opendev.org/712134 | 16:00 |
| *** bengates has quit IRC | 16:02 | |
| *** bengates has joined #openstack-kolla | 16:02 | |
| mgoddard | dking: btw, I don't read all IRC logs, so if you don't mention me by name I might miss your message | 16:03 |
| *** hongbin has joined #openstack-kolla | 16:05 | |
| mgoddard | dking: that is quite confusing | 16:05 |
| mgoddard | dking: have you discovered kayobe configuration dump? | 16:05 |
| *** bengates has quit IRC | 16:07 | |
| *** rpittau is now known as rpittau|afk | 16:07 | |
| dking | Thanks for the tip. I don't know that I've used that yet. I've just been dumping stuff from Ansible. | 16:08 |
| dking | I think it has something to do with those filters, but I don't understand why yet. | 16:08 |
| *** seco has quit IRC | 16:09 | |
| openstackgerrit | Merged openstack/kolla master: Zun: add CNI loopback binary to the image https://review.opendev.org/721044 | 16:10 |
| *** Luzi has quit IRC | 16:15 | |
| dking | Oh, it was my fault! I must have reverted things somewhere. I ended up setting external_net_name as a list rather than actually setting external_net_names. | 16:18 |
| openstackgerrit | Mark Goddard proposed openstack/kolla-ansible master: Remove support for Python 2 on the host https://review.opendev.org/713283 | 16:18 |
| openstackgerrit | Merged openstack/kolla-ansible master: Fix kibana deployment with openstack_cacert unset https://review.opendev.org/709091 | 16:19 |
| openstackgerrit | Merged openstack/kayobe stable/train: Use upper constraints when installing Tenks https://review.opendev.org/721082 | 16:20 |
| dking | Now, I'm receiving an error about kolla_internal_vip_address being invalid for localhost. It seems that internal_net_name is empty (""), even though networks.yml has: public_net_name: external | 16:29 |
| dking | ...I meant, I also have: internal_net_name: cloud | 16:30 |
| dking | mgoddard: Does networks.yml get picked up for localhost also? | 16:34 |
| *** evrardjp has quit IRC | 16:35 | |
| *** evrardjp has joined #openstack-kolla | 16:35 | |
| *** jbadiapa has quit IRC | 16:37 | |
| openstackgerrit | James Kirsch proposed openstack/kolla master: Add stunnel container for Glance service https://review.opendev.org/720850 | 16:41 |
| *** cah_link has quit IRC | 16:45 | |
| *** k_mouza has quit IRC | 16:46 | |
| *** gfidente is now known as gfidente|afk | 16:53 | |
| *** kklimonda has quit IRC | 16:54 | |
| *** kklimonda has joined #openstack-kolla | 16:55 | |
| openstackgerrit | Merged openstack/kolla-ansible master: Remove support for CentOS 7 https://review.opendev.org/713282 | 16:57 |
| sean-k-mooney | mgoddard: quick question. i have just deployed an all in one deployment with train-centos8 using master kolla-ansible and it worked fine. but i was wondering if i want to enable ironic can i add it to the same node and have both nova/libvirt and ironic deployed on the same host | 17:00 |
| sean-k-mooney | im not sure if i want/need ironic in my home deployment but it would be nice to be able to proviosu my other 4 servers too even though i will have those off 99% of the time | 17:02 |
| sean-k-mooney | so i was just wondering if kolla support colocating nova-ironic and nova-libvirt on the same system | 17:04 |
| sean-k-mooney | if not ill just create a vm and install ironic in that if i need it in the future | 17:05 |
| *** seco has joined #openstack-kolla | 17:06 | |
| *** e0ne has joined #openstack-kolla | 17:07 | |
| *** seco has quit IRC | 17:12 | |
| *** e0ne has quit IRC | 17:16 | |
| *** priteau has quit IRC | 17:19 | |
| *** ricolin has quit IRC | 17:29 | |
| *** portdirect has quit IRC | 17:30 | |
| *** portdirect has joined #openstack-kolla | 17:30 | |
| yoctozepto | sean-k-mooney: yes, nova-compute-ironic will be deployed on controllers by default | 17:35 |
| sean-k-mooney | yoctozepto: right but in a deployment with only 1 host can you deploy two nova compte conatienr on the same host with different config | 17:37 |
| sean-k-mooney | one useing ironic virt driver an the other using libvirt | 17:37 |
| sean-k-mooney | when i tried to do this 2-3 release ago you could not | 17:37 |
| sean-k-mooney | e.g. you need a minium of 2 hosts if you wanted 2 virt dirvers | 17:37 |
| sean-k-mooney | deploying the ironic serivce will work but i dont think kolla-ansible has support for deploying both compute agents on the same host | 17:38 |
| yoctozepto | sean-k-mooney: our CI does it so it should work | 17:38 |
| yoctozepto | it runs both libvirt and ironic on same nodes | 17:38 |
| sean-k-mooney | so a singel vm with both nova libvirt and nova ironic deployed | 17:39 |
| sean-k-mooney | ok i might try that | 17:39 |
| yoctozepto | ironic has different name | 17:39 |
| yoctozepto | host = primary for libvirt | 17:40 |
| yoctozepto | host = primary-ironic for ironic | 17:40 |
| sean-k-mooney | so i need to map both hostnames to the same host | 17:40 |
| yoctozepto | and they are discovered on deploy undert those different names | 17:40 |
| yoctozepto | no, no need to | 17:40 |
| sean-k-mooney | can you point me to the job defintion so i can see what is added to the global.yaml and inventory | 17:41 |
| sean-k-mooney | i could not find any docs for how to do this | 17:41 |
| yoctozepto | sean-k-mooney: docs should be enough but here you go: https://opendev.org/openstack/kolla-ansible/src/commit/9f54c6fd1b9b25e052c48a0dec90d38ff29cbcae/tests/templates/globals-default.j2#L81-L84 | 17:43 |
| sean-k-mooney | yoctozepto the ironci docs dont cover colocation https://docs.openstack.org/kolla-ansible/latest/reference/bare-metal/ironic-guide.html | 17:44 |
| *** evrardjp has quit IRC | 17:44 | |
| sean-k-mooney | when you set "enable_ironic: yes" in the past it prevented you form deploying libvirt | 17:44 |
| yoctozepto | sean-k-mooney: because it's by default | 17:44 |
| yoctozepto | then it has changed in the meantime | 17:44 |
| sean-k-mooney | yep as i said its been about 3 releases since i tried | 17:45 |
| yoctozepto | could use doc update most likely | 17:45 |
| openstackgerrit | James Kirsch proposed openstack/kolla-ansible master: Add support for encrypting glance/heat api https://review.opendev.org/707131 | 17:45 |
| *** seco has joined #openstack-kolla | 17:45 | |
| *** evrardjp has joined #openstack-kolla | 17:49 | |
| *** seco has quit IRC | 17:50 | |
| mgoddard | hi sean-k-mooney, as yoctozepto says it should work. The reason we don't document it is that it only really applies to the all-in-one case | 17:52 |
| sean-k-mooney | yep i have one server that is a tower server that is pretty power effeicnt and quite so i leave that running | 17:53 |
| sean-k-mooney | but i have 4 other servers taht are all in a single 2u rack mount chassis that i use for sriov work | 17:54 |
| sean-k-mooney | so im trying to confiur it so my tower server which will run my dev vms can optionally use ironic to provision my other servers | 17:54 |
| mgoddard | that should work just fine | 17:55 |
| sean-k-mooney | it looks like i just missed this being added | 17:55 |
| *** happyhemant has quit IRC | 17:55 | |
| sean-k-mooney | it predates teh refactor to launching containers via the role handles | 17:55 |
| sean-k-mooney | last time i tried was Newton i belive and it had issue back then. | 17:55 |
| sean-k-mooney | anyway thanks :) | 17:56 |
| sean-k-mooney | unfortuetly i guess kolla ansible droped the requiremetns for docs for all new features/roles at some point as it seam like it can do much more then is documented | 17:57 |
| sean-k-mooney | by the way that is not intended as a dig, i just noticed that as i try to enable service like octavia there are not docs for it and the same is true of many of the services. some just work but often you need to read the ansible playbooks to figure out how it works. | 18:02 |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Cleanup py27 support https://review.opendev.org/720764 | 18:02 |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Cleanup py27 support https://review.opendev.org/720764 | 18:04 |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Remove support for CentOS 7 and Python 2 https://review.opendev.org/720818 | 18:09 |
| *** hongbin has quit IRC | 18:21 | |
| *** seco has joined #openstack-kolla | 18:22 | |
| *** hongbin has joined #openstack-kolla | 18:26 | |
| *** hongbin has quit IRC | 18:27 | |
| *** seco has quit IRC | 18:27 | |
| *** cah_link has joined #openstack-kolla | 18:42 | |
| *** k_mouza has joined #openstack-kolla | 18:48 | |
| hrw | stable/train with official centos mirrors build goes quite nice | 18:50 |
| *** k_mouza has quit IRC | 18:52 | |
| *** seco has joined #openstack-kolla | 18:54 | |
| Fl1nt | @mgoddard, oh neat! thanks a lot, I'll definitely switch to venv. | 18:58 |
| *** Fl1nt has quit IRC | 18:58 | |
| sean-k-mooney | yoctozepto: mgoddard as an fyi it looks like there is some issue with ironic and iptables on centos 8 http://paste.openstack.org/show/792432/ | 18:59 |
| sean-k-mooney | if i run the command myself i get the same output | 19:00 |
| sean-k-mooney | sudo iptables --flush ironic-inspector | 19:00 |
| sean-k-mooney | iptables: Operation not supported. | 19:00 |
| sean-k-mooney | whats a little bit strange is --flush is supported but i do know that iptables/nftables implemenation changed between centos 7 a and 8 so maybe there is a step i missed that is required to make that work | 19:02 |
| openstackgerrit | Marcin Juszkiewicz proposed openstack/kolla stable/train: [WIP] switch to official CentOS 8 repositories https://review.opendev.org/721329 | 19:02 |
| hrw | we still lack centos-release-* packages to enable those repos but we no longer use delorean | 19:03 |
| * hrw off | 19:04 | |
| sean-k-mooney | imguessing centos 8 is default to iptables-nft but we need iptables-legacy | 19:05 |
| sean-k-mooney | ya so on the host centos 8 is using the nf tables version iptables v1.8.2 (nf_tables) | 19:10 |
| sean-k-mooney | it looks like the issue is related to https://opendev.org/openstack/kolla-ansible/commit/86e83faeb1fd088d44c5108a5ec835eba6316b2d | 19:20 |
| *** hjensas has joined #openstack-kolla | 19:23 | |
| *** e0ne has joined #openstack-kolla | 19:35 | |
| *** dciabrin has quit IRC | 19:37 | |
| *** dciabrin has joined #openstack-kolla | 19:37 | |
| sean-k-mooney | mgoddard: ok so if i remove the deploy.yml change introduced in ^ i think i can deploy ironic properly on centos-8 its passed where its was failing but is still running | 19:38 |
| sean-k-mooney | given that was added for upgrading betwen using iptables to dnsmasq can it now be removed? | 19:38 |
| *** e0ne has quit IRC | 19:40 | |
| *** e0ne has joined #openstack-kolla | 19:41 | |
| *** e0ne has quit IRC | 19:46 | |
| sean-k-mooney | reported as a bug here https://bugs.launchpad.net/kolla-ansible/+bug/1873951 | 19:49 |
| openstack | Launchpad bug 1873951 in kolla-ansible "ironinc inspector deploy fails due to iptables --flush" [Undecided,New] | 19:49 |
| r3ap3r | sean-k-mooney: do you happen to be using CentOS 8 Stream or CentOS 8.1? | 19:50 |
| sean-k-mooney | centos 8 stream | 19:50 |
| sean-k-mooney | since that should most closly track the next version fo rhel | 19:50 |
| sean-k-mooney | ill actully update the bug with that just incase there is a difference | 19:53 |
| r3ap3r | Ok, thanks. Just trying to get an idea of what people are using. From my understanding, there will be two different versions of RHEL and CentOS from here on out, Stream and non-Stream. Stream is meant to be semi-rolling release model the more closely aligns with Fedora while the "non-Stream" will stick with Major and Minor version releases. I may be mistaken but that is my current understanding. | 19:53 |
| sean-k-mooney | r3ap3r: well i normaly use ubunutu but since i work for redhat im at least trying to use centos | 19:56 |
| r3ap3r | sean-k-mooney: haha, understood. | 19:56 |
| *** seco has quit IRC | 20:00 | |
| r3ap3r | sean-k-mooney: for what it is worth, I have a C6145 that I am installing CentOS 8 on and have to use the drivers for the LSI 2008 from the dud elrepo to get them to work. If I am trying to install CentOS 8 Stream, it doesn't pick up the dud driver from the other USB drive I have plugged in but after flashing CentOS 8.1 it does pick up the dud drivers. Definitely a difference between the two. ;-) | 20:11 |
| *** k_mouza has joined #openstack-kolla | 20:15 | |
| r3ap3r | mgoddard: yoctozepto: do you happen to be running CentOS 8 Stream or CentOS 8.1 for testing your CI jobs if you don't mind me asking? | 20:16 |
| yoctozepto | last time I checked - stream | 20:18 |
| * yoctozepto goes to bed | 20:18 | |
| r3ap3r | Cool, thanks. | 20:19 |
| *** k_mouza has quit IRC | 20:20 | |
| sean-k-mooney | r3ap3r: oh i am not using the centos kernel i imendetly replaced it with the kernel-ml package from elrepo | 20:20 |
| sean-k-mooney | as i said i normally use ubuntu and in used to using ukuu to run the latest upstream longterm support kernel or upstream stable kernel | 20:21 |
| sean-k-mooney | i do not like distro patched kernels in general. i do use them but i have been bitten by disto kernels braking nested virt and dpdk in the past so i prefer to use a vanilla kernel where i can | 20:22 |
| r3ap3r | sean-k-mooney: Hmm, never had that problem with CentOS or Arch Linux but hey, there is a first time for everything. Thanks for the info, I always love learning how other people do things so I can better my processes. | 20:25 |
| sean-k-mooney | i have had mixed success with centos and nested virt | 20:25 |
| sean-k-mooney | i dislike that centos 8 was based on 4.18 when nested virt was enable upstream by defaul in linux in 4.19 | 20:26 |
| sean-k-mooney | many of the nested virt change have been backported but iw ould have prefered it to be based on 4.19 or 5.0 | 20:27 |
| r3ap3r | I have been using oVirt for the past year but decided to try my hand at OpenStack which led me here. I like Openstack better than oVirt for building training environments. Ah, I see what you are saying. That does make sense, missed it by one minor version update. | 20:28 |
| sean-k-mooney | yep so when rhel 8.0 lanuched or not everything had been backported. in 8.1 its more or less fine bar some edgecase that will be adress in 8.2 related to secure boot | 20:29 |
| sean-k-mooney | but as a result i like to track the latest sable upstream kernel to get those fixes as soon as they are merged and released | 20:30 |
| r3ap3r | I've always been curious about utilizing the latest stable upstream kernel on CentOS but a little hesitant due to "unexpected" results from using the "non-standard" kernel but I've been thinking about giving it a go due to moving to OpenStack. | 20:32 |
| sean-k-mooney | ya i should point out this is for my dev servers | 20:34 |
| sean-k-mooney | i work on nova mainly and do most of my dev on a openstack deployed with kolla | 20:34 |
| sean-k-mooney | so while i want it to be stable for my own work i am the only user of that install | 20:35 |
| sean-k-mooney | that said i have never really had issue on ubuntu with ukuu | 20:35 |
| r3ap3r | That is cool. I like to live a little dangerous so I may roll my production that way once I get my deployment process down. | 20:36 |
| sean-k-mooney | hehe well as long as you are prepared to fix it if it explodes. one of the benifts of ukuu that elrepo does not provide is all the upstream kerenls. e.g. you can pick to saty on jsut eh 5.4.x kernel with elrepo | 20:37 |
| sean-k-mooney | ukuu if you are not familar with it is the ubuntu kernel update utility and automate installing kerenls form the ubuntu-mainline ppa which has ever point release of the kernels avaiable | 20:38 |
| r3ap3r | My main distro everywhere but my servers is Arch Linux so I am well prepared for explosions, breakage and everything in between. ;-D | 20:39 |
| r3ap3r | I even thought about rolling OpenStack on Arch but I have to learn OpenStack better before I try that. lol | 20:39 |
| sean-k-mooney | i tried manjaro for a while and ya the same tought crossed my mind. i was suprissed with how outdate the pacakges were for arch | 20:40 |
| sean-k-mooney | havent tried vanilla arch but i might at somepoint | 20:41 |
| r3ap3r | Yeah, Manjaro is supposed to be "user friendly" Arch but I don't care for it. I have had Manjaro break on me 10 times more than Vanilla Arch. | 20:43 |
| r3ap3r | Started on Antergos before they retired the project and then went the Arch Way and haven't looked back. | 20:44 |
| r3ap3r | Arch Linux + i3 + Vim = productivity for me. | 20:44 |
| sean-k-mooney | :) unfortunetly devstack and kolla dont support arch so i wont be using it anytime soon form my day to day dev. since my laptop has an nvida gpu i just stick with popos on that front too since it more or less just works with nvida | 20:46 |
| *** cah_link has quit IRC | 20:47 | |
| r3ap3r | Not intending to start a "flame war" but I'm an AMD guy myself and ever since the opensourced pretty much every aspect of their drivers for their GPU's, it has made the AMD GPU support in Linux worlds better in my opinion. | 20:48 |
| sean-k-mooney | oh ya if you are on lunux you should use amd but we had a limited choice of laptops and this was the only option with a gpu | 20:50 |
| sean-k-mooney | well dedicated gpu | 20:50 |
| r3ap3r | My mindset is if it works, it works, use whatever you need to get the job done. At the end of the day, everything is a tool for us to use to accomplish our goal/task so use what works for YOU. | 20:51 |
| sean-k-mooney | yep | 20:51 |
| r3ap3r | Very true, Intel/Nvidia do tend to jump into every retail as quickly as possible. | 20:52 |
| openstackgerrit | James Kirsch proposed openstack/kolla master: Add stunnel container for Glance service https://review.opendev.org/720850 | 20:52 |
| *** e0ne has joined #openstack-kolla | 20:54 | |
| *** jimcrowleyibm has quit IRC | 20:59 | |
| *** jimcrowleyibm has joined #openstack-kolla | 20:59 | |
| *** jimcrowleyibm has quit IRC | 21:04 | |
| *** mcnugit has joined #openstack-kolla | 21:16 | |
| *** e0ne has quit IRC | 21:42 | |
| *** dosaboy has quit IRC | 21:46 | |
| *** dosaboy has joined #openstack-kolla | 22:02 | |
| *** e0ne has joined #openstack-kolla | 22:06 | |
| *** mcnugit has quit IRC | 22:12 | |
| *** born2bake has quit IRC | 22:33 | |
| *** TrevorV has quit IRC | 22:42 | |
| *** threestrands has joined #openstack-kolla | 23:18 | |
| *** hjensas has quit IRC | 23:49 | |
| *** hjensas has joined #openstack-kolla | 23:51 | |
| openstackgerrit | James Kirsch proposed openstack/kolla-ansible master: Add support for encrypting glance/heat api https://review.opendev.org/707131 | 23:53 |
| *** e0ne has quit IRC | 23:58 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!