| noonedeadpunk | mornings | 07:38 |
|---|---|---|
| noonedeadpunk | Moha: that really depends on the driver for cinder-volume. There're 2 modes cinder-volume can work in - active/active and active/passive. So you should check what mode your dirver can operate in | 07:46 |
| lsudre | Hello, after setting up a test OpenStack environment on VMs (on Proxmox), I am trying to understand how openstack-ansible works for a future deployment. I find it challenging to locate up-to-date resources on this topic. I am using Ubuntu 22.04 VMs, and the netplan config file from the openstack-ansible repo doesn't seem to make the required bridges (br-mgmt, br-vxlan, br-vlan, br-storage) function correctly. Do you have | 08:02 |
| lsudre | * correctly create a network compatible with OpenStack? A step-by-step tutorial would be amazing. Thank you. | 08:03 |
| noonedeadpunk | lsudre: so, before Antelope we require to create these bridges as pre-deployment step | 08:17 |
| noonedeadpunk | Starting from Antelope we do also include systemd-networkd role that can configure bridges automatically | 08:17 |
| hamidlotfi_ | hello, | 08:18 |
| hamidlotfi_ | That's good news @noonedeadpunk | 08:18 |
| noonedeadpunk | example on how to do so included here: https://docs.openstack.org/openstack-ansible/latest/user/network-arch/example.html#configuring-network-interfaces | 08:18 |
| noonedeadpunk | bridges will be created during openstack_hosts role (which is part of setup-hosts) | 08:19 |
| noonedeadpunk | lsudre: but you can still create networking manually with netplan, and example of netplan configuration could be found here: https://opendev.org/openstack/openstack-ansible/src/branch/master/etc/netplan | 08:20 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_horizon master: Fix linters and metadata https://review.opendev.org/c/openstack/openstack-ansible-os_horizon/+/888667 | 08:26 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_swift master: Fix linters and metadata https://review.opendev.org/c/openstack/openstack-ansible-os_swift/+/888672 | 09:45 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-hardening master: Fix linters and metadata https://review.opendev.org/c/openstack/ansible-hardening/+/888682 | 12:25 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-plugins master: Fix linters and metadata https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/888684 | 12:44 |
| NeilHanlon | o/ morning folks | 13:10 |
| NeilHanlon | (afternoon... i guess. for most ;) ) | 13:10 |
| noonedeadpunk | o/ | 13:32 |
| * noonedeadpunk still didn't have time to look at OVN thing, but it's still an issue | 13:32 | |
| * NeilHanlon either but it's on his list for this morning | 13:32 | |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_octavia master: Fix linters and metadata https://review.opendev.org/c/openstack/openstack-ansible-os_octavia/+/888687 | 13:38 |
| hamburgler | OVN thing - is this the inactivity probe? or something else? | 13:53 |
| mgariepy | hamburgler, i think this is the issue : https://zuul.opendev.org/t/openstack/build/e74702ad66d34e38a5d790a5d1aab9cc/log/job-output.txt#17507 | 13:58 |
| noonedeadpunk | hamburgler: it's smth regarding packaging issues or some requirements missmatch on rocky | 14:00 |
| hamburgler | ah gotcha :) ty. - had searched history for OVN and saw the inactivity probe timeout - https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/887192 - we ended up setting this also to 60000 but also happy to see it patched, noticed that it doesn't look like the value gets set on the compute node side though? - for example ovs-vsctl -v set manager "ptcp:6640:127.0.0.1" inactivity_probe=60000 (pr | 14:05 |
| hamburgler | seems to cover nb/sb only) | 14:05 |
| anskiy | hamburgler: yeah, I've only had problems with connection between Neutron and OVN NB, actually. | 14:06 |
| hamburgler | anskiy: specific version(s)? or any you've worked with because of the value? | 14:08 |
| anskiy | hamburgler: you mean, why I've choosed 60000 as default or?.. | 14:10 |
| hamburgler | oh no sorry, I mean if there was another connectivity issue :) | 14:11 |
| anskiy | oh, no, only this. But I do actually read somewhere while trying to fix this thing, that there're a couple of other timeouts, yes | 14:12 |
| anskiy | I've found 60000 in kolla-ansible's change in gerrit, so that looked kinda reasonable to me. I have no idea on which amount of instances/SGs this is gonna break again | 14:13 |
| anskiy | regarding setting this thing in OVS/ovn-controller -- I think this could only be needed to change in case of, maybe, super big compute nodes, like, maybe thousands of VMs?.. | 14:16 |
| anskiy | hamburgler: have you stumbled upon specific issue, that forced you to change this thing inv OVS? | 14:17 |
| hamburgler | When we set this last week - https://www.openvswitch.org/support/ovscon2019/day1/1436-OVSCON-Nouman.pdf it looks like 60000 should be desirable | 14:17 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_nova master: Fix linters and metadata https://review.opendev.org/c/openstack/openstack-ansible-os_nova/+/888692 | 14:17 |
| hamburgler | not a specific issue, we were just looking through logs and looked into it | 14:18 |
| hamburgler | I suppose there is some room for interpretation with that doc | 14:18 |
| anskiy | well, I wonder for how many inances/sgs this tuning is for... | 14:20 |
| hamburgler | probably more than my team is running at the moment :D | 14:21 |
| TadiosAbebe | Hello everyone, can anybody here tell me what the status of skyline dashboard with respect to openstack ansible is? couldn't find much when googling. | 16:09 |
| noonedeadpunk | TadiosAbebe: it's WIP I would say | 16:10 |
| noonedeadpunk | We have a role, that worked one day, we have a patch which would integrate the role, but it was never released yet for real | 16:11 |
| noonedeadpunk | we're aiming to get role in for the Bobcat release | 16:11 |
| noonedeadpunk | TadiosAbebe: the patch is https://review.opendev.org/c/openstack/openstack-ansible/+/859446 | 16:12 |
| noonedeadpunk | but it would need some more work/love before considering it as ready for production I would say | 16:12 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron master: Fix linters and metadata https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/888729 | 16:17 |
| TadiosAbebe | noonedeadpunk ohh great.Thank you very much, and i couldn't find a good example of deploying openstack with ovn using openstack ansible, the examples of openstack user config and user variables in the openstack ansible repositories still use linux bridges, any resources related to this? | 16:18 |
| noonedeadpunk | TadiosAbebe: I think this is the doc for OVN: https://docs.openstack.org/openstack-ansible-os_neutron/latest/app-ovn.html | 16:19 |
| noonedeadpunk | but yeah, we have updating default docs in our to-do list, though very little time to really update them | 16:19 |
| noonedeadpunk | Well, it seems I'm done with patches for new linter | 16:24 |
| noonedeadpunk | Now only left to unblock gates | 16:24 |
| TadiosAbebe | i don't know why i haven't seen this page before but great i will look in to it. the thing is i have been trying to deploy openstack for the past about 4 month for production environment, i have been getting better but recently i got stuck on after deployment. the problem being instances outside connectivity not working. i know this is a problem with my config files but i still haven't figured out what. I'll read the link you provided and | 16:26 |
| TadiosAbebe | get back thank you very much for your response | 16:26 |
| noonedeadpunk | and you're using OVN, right? | 16:28 |
| TadiosAbebe | that is correct | 16:29 |
| noonedeadpunk | well, I'd suggest pinging jamesdenton or maybe mgariepy will be able to help you out. As I personally don't use OVN anywhere | 16:29 |
| noonedeadpunk | I'd assume this is smth with gateway nodes, but have quite vague understanding | 16:30 |
| noonedeadpunk | Can help with OVS or LXB though :D | 16:30 |
| mgariepy | i have zed running fine with ovn + gateway chassis on my compute nodes. | 16:31 |
| TadiosAbebe | oh okay, the reason i am using ovn is that i read that linux bridges have been depreciated in the zed realeas or so.. | 16:31 |
| mgariepy | what is the issue exactly :) ? | 16:32 |
| TadiosAbebe | i am trying to deploy zed too but i think there is a problem with my configs | 16:32 |
| mgariepy | can you paste is somewhere? | 16:32 |
| hamburgler | noonedeadpunk: sorry if I am asking again :D, do you know when 27.0.1 will be released? | 16:33 |
| TadiosAbebe | definitely can i get back with you here tomorrow? I don't have my PC and work environment with me right now | 16:33 |
| noonedeadpunk | hamburgler: once https://review.opendev.org/c/openstack/releases/+/888117 will have second vote | 16:33 |
| noonedeadpunk | can be today or can be end of the week | 16:35 |
| hamburgler | thank you :) | 16:36 |
| mgariepy | TadiosAbebe, sure. i'm here pretty much all the time. | 16:38 |
| mgariepy | ping me so i can see the messages and i will respond. | 16:38 |
| TadiosAbebe | I have to learn how to ping someone in irc😀, i have recently encounter issues in my deployment with gnocchi and Dmitriy Rabotyagov suggest to reach you guys through this irc for further suggestions and help and pointed out that it is a very friendly community, and i am very pleased. Thank you very much | 16:41 |
| mgariepy | just add my username : mgariepy in your sentence ;p | 16:45 |
| TadiosAbebe | mgariepy: great | 16:47 |
| mgariepy | anyone use mig here ? | 17:53 |
| noonedeadpunk | we use regular vgpus | 17:54 |
| noonedeadpunk | btw, have you already spotted a deprecation notice on the latest vgpu drivers? | 17:55 |
| mgariepy | nop | 17:55 |
| noonedeadpunk | it's "fun" | 17:55 |
| mgariepy | regular vgpus with the fun version matching with the host and the vm ? | 17:55 |
| noonedeadpunk | Yes, it is | 17:56 |
| noonedeadpunk | is it different with MIG? | 17:56 |
| noonedeadpunk | But what I was talking about is https://docs.nvidia.com/grid/latest/whats-new-vgpu/index.html | 17:56 |
| noonedeadpunk | "Feature Support Withdrawn" section | 17:57 |
| mgariepy | afaik mig split the gpu then you can create vm with passthrough ? | 17:57 |
| noonedeadpunk | I think difference between mig and vgpu is only in memory sharing | 17:57 |
| noonedeadpunk | Like MIG has dedicated memory provisioning, vgpu shared | 17:57 |
| noonedeadpunk | they both represented as mediated devices (mdev) | 17:58 |
| noonedeadpunk | And both require pretty much same licensing | 17:58 |
| noonedeadpunk | well, mig has couple of extra sweet things, like splitting GPU into different supported flavors rather then requirement to split GPU with 1 flavor | 17:59 |
| mgariepy | ha i didn't knew it needed licence. | 17:59 |
| noonedeadpunk | on client side | 17:59 |
| mgariepy | anyway trtying hard not to have to work with that. | 17:59 |
| noonedeadpunk | otherwise it's not usable | 17:59 |
| noonedeadpunk | https://docs.nvidia.com/grid/13.0/grid-licensing-user-guide/index.html#software-enforcement-grid-licensing | 18:00 |
| noonedeadpunk | nvidia is "fun" overall | 18:02 |
| mgariepy | wow. | 18:03 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!