*** macz_ has quit IRC | 00:12 | |
*** tosky has quit IRC | 00:27 | |
*** spatel has quit IRC | 00:32 | |
*** cshen has quit IRC | 00:33 | |
openstackgerrit | Merged openstack/openstack-ansible stable/ussuri: Fix manila tempest jobs https://review.opendev.org/759426 | 01:05 |
---|---|---|
openstackgerrit | Merged openstack/openstack-ansible-os_aodh master: Reduce number of processes on small systems https://review.opendev.org/759883 | 01:40 |
*** PrinzElvis has quit IRC | 01:50 | |
*** sri_ has quit IRC | 01:50 | |
*** sri_ has joined #openstack-ansible | 01:51 | |
*** PrinzElvis has joined #openstack-ansible | 01:51 | |
*** gyee has quit IRC | 02:10 | |
*** MickyMan77 has joined #openstack-ansible | 02:26 | |
*** dave-mccowan has joined #openstack-ansible | 02:29 | |
*** MickyMan77 has quit IRC | 02:35 | |
*** MickyMan77 has joined #openstack-ansible | 03:07 | |
*** MickyMan77 has quit IRC | 03:16 | |
*** irclogbot_1 has quit IRC | 04:31 | |
*** irclogbot_3 has joined #openstack-ansible | 04:32 | |
*** MickyMan77 has joined #openstack-ansible | 04:45 | |
*** MickyMan77 has quit IRC | 04:54 | |
*** akahat is now known as akahat|PTO | 05:20 | |
*** evrardjp has quit IRC | 05:33 | |
*** evrardjp has joined #openstack-ansible | 05:33 | |
*** MickyMan77 has joined #openstack-ansible | 05:50 | |
*** MickyMan77 has quit IRC | 05:59 | |
*** gregwork has quit IRC | 06:05 | |
*** dmsimard has quit IRC | 06:48 | |
*** MickyMan77 has joined #openstack-ansible | 06:55 | |
*** shyamb has joined #openstack-ansible | 06:59 | |
*** MickyMan77 has quit IRC | 07:04 | |
*** miloa has joined #openstack-ansible | 07:15 | |
pto | jrosser: I would love to test the federated setup. I experience the same issue when I tried to deploy queens and the same workaround worked. | 07:21 |
*** rpittau|afk is now known as rpittau | 08:15 | |
*** macz_ has joined #openstack-ansible | 08:16 | |
*** macz_ has quit IRC | 08:20 | |
*** cshen has joined #openstack-ansible | 08:32 | |
noonedeadpunk | morning | 08:32 |
*** tosky has joined #openstack-ansible | 08:50 | |
*** sshnaidm|rover is now known as sshnaidm|roff | 09:06 | |
*** sshnaidm|roff is now known as sshnaidm|off | 09:06 | |
pto | I there any plans to release a new stable release/tag for ussuri soon? | 09:13 |
*** NewJorg has quit IRC | 09:25 | |
*** NewJorg has joined #openstack-ansible | 09:27 | |
*** yann-kaelig has joined #openstack-ansible | 09:31 | |
noonedeadpunk | sure | 09:41 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_magnum master: Drop magnum distro CI jobs https://review.opendev.org/760525 | 09:51 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_magnum stable/ussuri: Drop magnum distro CI jobs https://review.opendev.org/760527 | 09:51 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_magnum stable/ussuri: Add deployment of keystone_auth_default_policy https://review.opendev.org/759471 | 09:52 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_magnum stable/ussuri: Add deployment of keystone_auth_default_policy https://review.opendev.org/759471 | 09:52 |
*** gokhani has joined #openstack-ansible | 10:15 | |
openstackgerrit | Merged openstack/openstack-ansible-os_manila stable/ussuri: Add centos-8 support https://review.opendev.org/759396 | 10:52 |
evrardjp | hello folks | 11:00 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible stable/ussuri: Actually use ANSIBLE_FORKS in openstack-ansible.rc https://review.opendev.org/760366 | 11:01 |
evrardjp | I am working on ansible-keepalived, and removing a bunch of crap that we piled up over the years, now that packaging is better, and ansible is better. Would someone like to review? | 11:01 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible stable/stein: Actually use ANSIBLE_FORKS in openstack-ansible.rc https://review.opendev.org/760540 | 11:02 |
noonedeadpunk | evrardjp: o/ | 11:03 |
evrardjp | hello noonedeadpunk :) | 11:04 |
evrardjp | How is everyone? | 11:04 |
noonedeadpunk | I hope pretty good:) | 11:04 |
noonedeadpunk | moving slowly but surely:) | 11:04 |
noonedeadpunk | I can take a look if that helps:) | 11:05 |
evrardjp | Yeah, I think I would like that. I shall try again molecule, to see if I can have coverage, because right now, I am doing that on vagrant, and it's annoying nowadays :p | 11:06 |
evrardjp | haha ofc it would be better with the link: https://github.com/evrardjp/ansible-keepalived/pull/157/ | 11:06 |
evrardjp | I thought I already sent it. Where is my mind?! | 11:06 |
noonedeadpunk | I already found it :p | 11:06 |
evrardjp | you've inspired me to work on it :) | 11:07 |
noonedeadpunk | sorry I didn't mean it :p | 11:07 |
noonedeadpunk | Wasn't like going to push you to spend time on it, but it's great I did | 11:08 |
evrardjp | yeah I think it would be good to have proper coverage | 11:09 |
*** dmsimard has joined #openstack-ansible | 11:11 | |
noonedeadpunk | evrardjp: do you recal what was the {{ reason }} why we had `Prevent keepalived from starting on install` task at the first place? | 11:12 |
evrardjp | yeah, it was when a ppa was used, some ppa were just starting keepalived, and reloaded with a different configuration (that's bad), or started a broken keepalived on first install | 11:17 |
evrardjp | the only way to get through this would have been to ignore_errors but I didn't like it | 11:18 |
evrardjp | I don't regret that previous decision, but the packaging is now in a better place | 11:18 |
evrardjp | I probably should have explained that better in each of the commit messages of the PR | 11:19 |
noonedeadpunk | Ah, I never paid attention if keepalived is started by default currently, but it's not, so yeah - makes sense | 11:20 |
* noonedeadpunk just reailizied has no idea how to efficiently review on github | 11:22 | |
*** gshippey has joined #openstack-ansible | 11:29 | |
* evrardjp thinks there is no efficient way to review on github when you have tried gerrit :) | 11:34 | |
evrardjp | did OSA move away from the with_first_found already? | 11:35 |
* evrardjp opens up roles | 11:36 | |
noonedeadpunk | no | 11:36 |
noonedeadpunk | should we?:) | 11:36 |
evrardjp | IIRC, the with_ syntax was supposed to be deprecated at some point | 11:36 |
noonedeadpunk | (I might missed more efficient way) | 11:36 |
evrardjp | I find it less readable than with_ tbh | 11:37 |
noonedeadpunk | well, it's not and working with 2.10 | 11:37 |
evrardjp | I think Ansible rolled back on this then :) | 11:37 |
noonedeadpunk | I think they've rejected that idea moving to loops completely | 11:37 |
evrardjp | "We have not deprecated the use of with_<lookup> - that syntax will still be valid for the foreseeable future." | 11:38 |
evrardjp | that's good | 11:38 |
evrardjp | I guess I can keep mine then | 11:38 |
evrardjp | :D | 11:38 |
noonedeadpunk | btw, I'm trying to understand why in the world removal of mysql-common drops keepalived | 11:44 |
noonedeadpunk | and how to workaround that... | 11:45 |
mgariepy | morning everyone. | 11:48 |
noonedeadpunk | uh, it's requiement for keepalived 2.... | 11:48 |
noonedeadpunk | \o/ | 11:48 |
*** shyamb has quit IRC | 11:49 | |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-galera_server master: Do not remove mysql-common during upgrade https://review.opendev.org/760541 | 11:53 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Fix upgrade jobs for bind-to-mgmt https://review.opendev.org/758461 | 11:54 |
sauloasilva1 | Hi. I used to setup openstack using a custom scripts and now a I implemented using the openstack-ansible scripts . At my setup the log files are in separate files , with openstack-ansible it is journalctl log . How is the easiest way to debug a nova-compute and cinder problem ? | 12:11 |
noonedeadpunk | sauloasilva1: so if it's not aio deployment, then most likely cinder and nova-compute are on different hosts. Anyway you can get specific service logs with `journalctl -u nova-compute` (or whatever service you need) | 12:13 |
noonedeadpunk | you can use journalctl -f to follow log as well | 12:13 |
sauloasilva1 | OK . so what is the log server utilization , I tried to run journalctl -f on it and no log informatio about openstack at all | 12:17 |
noonedeadpunk | um, if you refer to rsyslog server, then I think this is kind of obsoleted. | 12:21 |
noonedeadpunk | logs are currently stored only on the hosts and containers, but it's pretty easy to | 12:21 |
noonedeadpunk | * to forward them to the log collector you have | 12:21 |
sauloasilva1 | OK , so that is not the default for all services . | 12:22 |
noonedeadpunk | here's sample for the graylog https://opendev.org/openstack/openstack-ansible-ops/src/branch/master/graylog/graylog-forward-logs.yml | 12:22 |
sauloasilva1 | I got it at containers . | 12:22 |
noonedeadpunk | by default log is stored in journalctl on computes, controllers and containers and it's not forwarded to any log collector | 12:22 |
*** rh-jelabarre has joined #openstack-ansible | 12:27 | |
*** MickyMan77 has joined #openstack-ansible | 12:34 | |
*** djhankb has quit IRC | 12:37 | |
*** djhankb has joined #openstack-ansible | 12:38 | |
*** rfolco has joined #openstack-ansible | 12:41 | |
*** MickyMan77 has quit IRC | 12:43 | |
*** spatel has joined #openstack-ansible | 12:56 | |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_neutron master: Fix OVN functional test https://review.opendev.org/760188 | 12:58 |
*** lkoranda has joined #openstack-ansible | 12:59 | |
pto | Is it supported to use ansible-vault with user_secrets.yml? | 13:01 |
noonedeadpunk | Not sure if anyone tried but I think it should be | 13:06 |
noonedeadpunk | eventually user_secrets.yml is just variables file which is loaded with `-e user_secrets.yml` | 13:07 |
noonedeadpunk | since ansible-vault will get variables defined - it should work | 13:07 |
*** gokhani has quit IRC | 13:09 | |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/ansible-role-systemd_networkd master: Updated from OpenStack Ansible Tests https://review.opendev.org/754195 | 13:10 |
*** MickyMan77 has joined #openstack-ansible | 13:18 | |
*** MickyMan77 has quit IRC | 13:26 | |
*** pto has quit IRC | 13:32 | |
*** pto_ has joined #openstack-ansible | 13:32 | |
*** yann-kaelig has quit IRC | 13:33 | |
openstackgerrit | Merged openstack/openstack-ansible-galera_server stable/stein: Bump galera version https://review.opendev.org/758462 | 13:38 |
evrardjp | pto_: I did it long ago but it is _terribly_ annoying. | 13:40 |
*** pto_ is now known as pto | 13:43 | |
pto | evrardjp: Thanks for clarifying this. I think i will make another solution with scripts instead | 13:43 |
evrardjp | I am generally worried when someone starts a sentence with _make another solution_ :) | 13:44 |
evrardjp | I think it's fine to use ansible-vault, and you should probably use that. | 13:45 |
evrardjp | But I don't see a reason why we couldn't have your own wrapper, doing the same thing as openstack-ansible CLI, by adding your secret sauce | 13:45 |
evrardjp | you couldn't* | 13:45 |
*** jamesdenton has joined #openstack-ansible | 13:46 | |
evrardjp | it's just that you will need to maintain it, and it's probably a hassle for very little results: The variables kept in secret are mostly exposed in some other way. | 13:46 |
mgariepy | yes it works i use it. | 13:46 |
evrardjp | mgariepy: do ppl still have to type their vault password at every single playbook nowadays? | 13:47 |
mgariepy | i do use gpg agent for the vaultpass | 13:47 |
mgariepy | --vault-password-file=/root/vaultpass.sh | 13:47 |
evrardjp | I see | 13:47 |
mgariepy | witch call `gpg --batch --use-agent --decrypt /root/vault_passphrase.gpg` | 13:48 |
evrardjp | isn't it painful for your colleagues? :p | 13:48 |
evrardjp | (or for your CI) | 13:48 |
mgariepy | the gpg key is on the server. | 13:48 |
evrardjp | I imagine how painful the loss of the gpg key could be :D | 13:49 |
mgariepy | i would put the prod password in a CI anyway ;p | 13:49 |
evrardjp | hahah | 13:49 |
mgariepy | and i don't have a CI yet for that.. | 13:50 |
evrardjp | I guess no choice | 13:50 |
evrardjp | I don't judge :) | 13:50 |
mgariepy | we use https://www.passwordstore.org/ to share password and stuff. | 13:50 |
evrardjp | I love passwordstore! | 13:50 |
masterpe | I'm looking in designate and with powerdns 4 as dns servers but I need to fill in at the targets.masters.host the designate containers. But in the example https://docs.openstack.org/designate/train/admin/backends/pdns4.html is only specify's one ip address | 13:51 |
mgariepy | yep password store mostly works 100% of the time ahha | 13:51 |
mgariepy | i found a bug with xclip | 13:51 |
mgariepy | ahha | 13:51 |
evrardjp | hahah | 13:52 |
evrardjp | did you try the team based one? Or just the standard one | 13:52 |
noonedeadpunk | masterpe: can share my config for powerdns if interested | 13:52 |
masterpe | But how do I specify all three designate containers? With host: "{{ groups['designate_all'] | default([]) }}" ???? | 13:52 |
mgariepy | we share the password with team | 13:52 |
masterpe | noonedeadpunk: would be great | 13:52 |
mgariepy | we have multiple passwrod repo in fact. since i'm in multiple teams :D | 13:53 |
evrardjp | I see | 13:53 |
mgariepy | the things i love about it is that it's 1 simple and not too subject to loosing access because the compagny X got issues. | 13:54 |
noonedeadpunk | masterpe: http://paste.openstack.org/show/799545/ | 13:55 |
mgariepy | i don't like these cloud based password service lol | 13:55 |
masterpe | noonedeadpunk: thank that helps alot | 13:56 |
evrardjp | mgariepy: we neither! | 13:56 |
evrardjp | how is everyone btw? | 13:57 |
mgariepy | doing good. playbing with ovn now :D | 13:57 |
noonedeadpunk | masterpe: and ns1.domain.com is resolved to ip.ad.re.ss1 and ns2.domain.com to ip.ad.re.ss2 in my case | 13:57 |
masterpe | noonedeadpunk: Am I correct that the documentation is incorrect: https://docs.openstack.org/designate/train/admin/backends/pdns4.html | 13:57 |
mgariepy | how are you evrardjp ? | 13:58 |
masterpe | noonedeadpunk: I thould it would be somethink link that. | 13:58 |
evrardjp | busy, busy.... but it's friday, so I am taking my time :p | 13:58 |
mgariepy | lol | 13:58 |
evrardjp | for once | 13:58 |
noonedeadpunk | well, you need to take some time off let's say. at least once :) | 13:59 |
evrardjp | I am happy to read familiar names in this chan, it's nice :) | 13:59 |
mgariepy | are you still doing some OSA stuff ? | 13:59 |
evrardjp | not at all | 13:59 |
*** MickyMan77 has joined #openstack-ansible | 13:59 | |
evrardjp | sadly | 13:59 |
mgariepy | what are you doing these day ? | 13:59 |
evrardjp | I miss you all :) | 13:59 |
jamesdenton | you are missed | 14:00 |
noonedeadpunk | can't really express how we are missig you | 14:00 |
evrardjp | well, it involves java, so I am not sure you really want to hear about my work :p | 14:00 |
jamesdenton | the stuff nightmares are made from | 14:01 |
mgariepy | indeed no java for me ;p | 14:01 |
evrardjp | haha | 14:01 |
evrardjp | jamesdenton: I see you know what you are talking about. | 14:01 |
* noonedeadpunk goes to get moderator permissions to ban for java xD | 14:01 | |
mgariepy | you should be using a lot of memory.. lo\ | 14:01 |
jamesdenton | we had a thing a long time ago. | 14:01 |
evrardjp | TBH I was pleasantly surprised of some recently made optimizations | 14:01 |
evrardjp | you could have a java application server in 30 MB of ram | 14:02 |
evrardjp | in a container! | 14:02 |
jamesdenton | s/MB/GB | 14:02 |
noonedeadpunk | is it hello world or some real application? | 14:02 |
evrardjp | hahaha | 14:02 |
*** lkoranda has quit IRC | 14:02 | |
*** lkoranda has joined #openstack-ansible | 14:03 | |
evrardjp | noonedeadpunk: haha fair question. It's pretty much hello world, I mean it's something without JCache already :) | 14:03 |
noonedeadpunk | as if it is I should consider looking into it one day again | 14:03 |
evrardjp | well, it's still _java_ | 14:03 |
evrardjp | hahaha | 14:03 |
evrardjp | mgariepy: I suppose everyone is playing with OVN now? | 14:03 |
evrardjp | did neutron made the final move to make it standard? | 14:04 |
noonedeadpunk | masterpe: well I don't think it's incorrect, just incomplete maybe. as I think they suppose that pdns will run on the same host as designate which is not the case in real world | 14:04 |
jamesdenton | it's in-progress | 14:04 |
*** MickyMan77 has quit IRC | 14:04 | |
* noonedeadpunk still likes lxb.... | 14:05 | |
*** MickyMan77 has joined #openstack-ansible | 14:05 | |
noonedeadpunk | probably I just don't have a usecase why I'd need ovn or ovs... | 14:07 |
jamesdenton | why complicate things if you don't have to | 14:07 |
masterpe | noonedeadpunk: how do you secure designate-mdns there I could not find a setting where I can specify the AXFR traffic, from where I should come from. | 14:08 |
evrardjp | jamesdenton: job security ? | 14:08 |
masterpe | Only possibilty that I see or a firewall or a desiganted network. | 14:08 |
evrardjp | jamesdenton: I am not for that, don't get me wrong :) | 14:08 |
jamesdenton | :) | 14:09 |
noonedeadpunk | masterpe: I have a standalone network for the designate | 14:09 |
*** MickyMan77 has quit IRC | 14:09 | |
noonedeadpunk | I did smth like this, not sure if it needs update https://docs.openstack.org/openstack-ansible/latest/reference/configuration/extra-networks.html | 14:10 |
evrardjp | has anyone toyed with molecule in here? Just curious | 14:12 |
mgariepy | i want ovn because lxb works wonderfully until you have 500 networks and a network node dies. | 14:13 |
noonedeadpunk | jrosser: did, but it's still in poc | 14:13 |
mgariepy | then it does take forever to sync back. | 14:13 |
noonedeadpunk | evrardjp: oh, disregard and sorry jrosser for pinging you... I mixed up with mitogen... | 14:15 |
spatel | jamesdenton: morning, i am trying to install OSA with openvswitch in my lab so quick question on that i do need openswitch on controller node for br-vlan and br-vxlan right? (i can use br-mgmt in linuxbridge) | 14:16 |
evrardjp | oh yeah mitogen, that was long ago I heard that! :) | 14:17 |
spatel | I am playing with DPDK so need lab to play :) | 14:17 |
evrardjp | I remember experiments | 14:17 |
jamesdenton | spatel you can keep br-mgmt and the ovs playbooks will configure the provider bridge for you | 14:17 |
jamesdenton | just read thru the docs a few times: https://docs.openstack.org/openstack-ansible-os_neutron/latest/app-openvswitch.html | 14:18 |
spatel | reading... | 14:18 |
jamesdenton | container_bridge will be the provider bridge name (br-provider) and network_interface, if defined, will be the interface connected to that bridge for you | 14:19 |
spatel | hmm.. ok let me finish that article and come back here. ( how do i configure bonding? in openvswitch or regular linux way) | 14:21 |
*** cshen has quit IRC | 14:21 | |
jamesdenton | meh, i would stick to the regular way for now | 14:21 |
spatel | my goal is to do DPDK testing | 14:22 |
spatel | some folks doing bonding inside OVS | 14:22 |
jamesdenton | ok, well, there's a little different method for that | 14:23 |
jamesdenton | https://docs.openstack.org/openstack-ansible-os_neutron/latest/app-openvswitch-dpdk.html | 14:23 |
noonedeadpunk | jamesdenton: btw regarding ovn and neutron testing - it's really surprising for me how and why integrated test works... | 14:23 |
jamesdenton | yes, you can do bonding w/ OVS, it's just none of it is automated here. | 14:23 |
jamesdenton | noonedeadpunk i saw your comment. was gonna spin up lxc focal and see how and why it worked there but not on metal | 14:23 |
jamesdenton | just. waiting. for. git. clone. | 14:24 |
noonedeadpunk | eventually it's also lxc job | 14:24 |
* jamesdenton forgets to implement parallel patch | 14:24 | |
spatel | jamesdenton: thank for link, let me start and see how it goes. (otherwise i will keep asking you stupid question). | 14:25 |
jamesdenton | it's all good. i expect the docs could use some clarifications and updates, it's been a few cycles | 14:26 |
spatel | +1 | 14:27 |
*** d34dh0r53 has quit IRC | 14:28 | |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_nova master: Install qemu from buster-backports https://review.opendev.org/760567 | 14:29 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Bump SHAs for master https://review.opendev.org/755973 | 14:29 |
*** dave-mccowan has quit IRC | 14:33 | |
*** macz_ has joined #openstack-ansible | 14:35 | |
*** dave-mccowan has joined #openstack-ansible | 14:36 | |
*** d34dh0r53 has joined #openstack-ansible | 14:38 | |
jamesdenton | spatel what NIC do you plan to use? | 14:42 |
spatel | Intel Corporation 82599 10 Gigabit Dual Port | 14:43 |
jamesdenton | old school cool | 14:43 |
spatel | jamesdenton: :( | 14:44 |
spatel | jamesdenton: do you think DPDK will be supported on it? | 14:44 |
jamesdenton | i believe it is | 14:44 |
spatel | hope i get performance close to SRIOV | 14:44 |
spatel | i want to move away from sriov | 14:44 |
jamesdenton | i think you can get close. There's a lot of tunables that I'm not really familiar with | 14:45 |
spatel | can i run dpdk and sriov together under single cloud | 14:45 |
spatel | incase dpdk won't do good then i can turn on sriov | 14:46 |
jamesdenton | you can definitely have them both in the same cloud | 14:46 |
spatel | that would be good. | 14:47 |
spatel | linuxbridge is great but look like time to move forward and use next gen technology | 14:47 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible stable/stein: Actually use ANSIBLE_FORKS in openstack-ansible.rc https://review.opendev.org/760540 | 14:47 |
openstackgerrit | Merged openstack/openstack-ansible-galera_client stable/stein: Bump galera version https://review.opendev.org/758464 | 14:50 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_ironic master: Define condition for the first play host one time https://review.opendev.org/756340 | 14:51 |
miloa | Hi, we are installing Train on our dev, method distro, OS ubuntu. Seems un os_keystone that install of packet pyhton3-ldappool is missing ? | 14:53 |
spatel | jamesdenton: is this true in coming future community going to remove linux-bridge because they have no maintainer, and ovs will be default. | 14:56 |
evrardjp | spatel: not really | 14:57 |
evrardjp | I think the idea was to move everyone to OVN | 14:57 |
evrardjp | see a little above | 14:57 |
evrardjp | ovn does use ovs, but it's not like the current ovs vs linuxbridge way | 14:58 |
spatel | agreed, plan is to move on and use pure SDN | 14:59 |
evrardjp | but I am not really involved in this, better ask other ppl, I am has-been | 14:59 |
evrardjp | there were so many other ways for scaling taht didn't involve that... but meh. | 15:00 |
spatel | OVS was buggy couple years ago but its getting more and more mature so looking forward to use OVS, majority of cloud provide using OVS for default deployment. | 15:00 |
noonedeadpunk | spatel: everyone looking forward to move to ovn from ovs, so probably you should reconsider:) | 15:02 |
jamesdenton | spatel there were rumblings at a PTG maybe 6-12 months ago about deprecating LXB, but it's fairly heavily used, and there was some push back. I'm not really sure what the current sentiment is. It seems there is/has been some idea that there needed to be parity with LXB and OVS, but I don't think that's necessary. Anyway, I am fairly certain OVN will supplant ML2/OVS but as for LXB, i'm not sure. May never go away if | 15:03 |
jamesdenton | there's interested people (that are also interested in maintaining it) | 15:03 |
*** cshen has joined #openstack-ansible | 15:04 | |
spatel | I have zero experience running OVS so moving on OVN would be big deal for me in production. currently trying to understand how i can run OVS+DPDK to improve my workload. | 15:07 |
spatel | does OSA has full deployment support with OVN in CentOS environment? | 15:07 |
jamesdenton | well, you can add some ovs nodes with your existing lxb nodes if you want to do a 90/10 or 80/20 | 15:08 |
jamesdenton | we can deploy OVN on centos but i don't think OSA has anyone using OVN in prod (yet). So i don't know how resilient the control plane is based on the way we deploy it | 15:08 |
spatel | oh wait so you are saying I can run linuxbridge and ovs under same cloud deployment? | 15:09 |
mgariepy | it's defenitely missing ha stuff for ovn | 15:09 |
spatel | i have 300 compute nodes in production running LXB and i add 10 more compute nodes running OVS+DPDK? | 15:10 |
spatel | can i do that way jamesdenton? | 15:10 |
spatel | i thought its binary choice LXB or OVS, you can run both together (may be i am wrong) | 15:11 |
miloa | On same deployment (Ubuntu/Distro/Train) in /etc/ansible/roles/ceph_client/tasks/ceph_install_python_libs.yml it looks for site-package but in distro method the libs are in dist-package. Could you confirm ? | 15:17 |
jamesdenton | spatel same cloud, yes. same host, no | 15:18 |
spatel | I have 10 extra compute nodes can i install OVS on them and add into existing cloud? | 15:19 |
jamesdenton | spatel with a bit of inventory/group munging, you should be able to | 15:19 |
jamesdenton | i've done it, just don't recall the specifics | 15:20 |
spatel | jamesdenton: i totally understand hacking inventory etc.. all i need to see if that is possible then i can test my application in each type of networking | 15:20 |
noonedeadpunk | miloa: iirc yes. and for distro method ceph_install_python_libs is not needed as it does symlinking into venv, while for distro you won't have venv | 15:20 |
jamesdenton | that's a good plan | 15:21 |
spatel | i like that plan to run cloud in mix environment, let me work toward that direction :) | 15:22 |
jamesdenton | sounds good | 15:22 |
jamesdenton | i'll look for your blog on the subject soon :) | 15:22 |
spatel | oh yess :) | 15:23 |
miloa | noonedeadpunk: ok, but when i deploy using distro method, I've got an error on ceph_install_python_libs looking for lib in site-packages. | 15:23 |
noonedeadpunk | sound like a bug..... | 15:24 |
noonedeadpunk | well, yes, it is..... | 15:25 |
miloa | noonedeadpunk: in ceph_client the task ceph_install_python_libs is active if openstack_service_venv_bin != '' if we change it by when openstack_service_venv_bin is defined and ... !='' would it help ? | 15:32 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_glance master: Do not symlink ceph libraries for distro path https://review.opendev.org/760581 | 15:33 |
noonedeadpunk | miloa: ^ can you check this please? | 15:33 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_glance master: Do not symlink ceph libraries for distro path https://review.opendev.org/760581 | 15:33 |
noonedeadpunk | sorry updated version ^ | 15:33 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_cinder master: Do not symlink ceph libraries for distro path https://review.opendev.org/760582 | 15:35 |
miloa | Thanks :) I will test it rigth now on train :) I think that it could apply on nova, manilla, cinder and glance i think :) grep openstack_service_venv_bin -R /etc/ansible/roles/ | 15:37 |
noonedeadpunk | it is | 15:37 |
noonedeadpunk | and making patches for all of them | 15:37 |
miloa | Many thanks :) | 15:37 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_nova master: Do not symlink ceph libraries for distro path https://review.opendev.org/760583 | 15:38 |
mgariepy | jamesdenton, https://pastebin.com/AN9rQAvF network mapping stuff it not so good for containers ovn stuff :D | 15:39 |
jamesdenton | which container is that in? | 15:40 |
jamesdenton | and what does your bridge config look like? | 15:40 |
mgariepy | my ovn_gateway | 15:40 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_manila master: Do not symlink ceph libraries for distro path https://review.opendev.org/760585 | 15:41 |
jamesdenton | looks like it tried creating a bridge named eth11. | 15:41 |
mgariepy | https://github.com/openstack/openstack-ansible-os_neutron/blob/master/tasks/providers/setup_ovs_ovn.yml#L75 https://pastebin.com/CW71avr3 | 15:42 |
jamesdenton | yeahm oops | 15:43 |
jamesdenton | interface_mapping.split(':')[0] | 15:43 |
mgariepy | yep, on container but not on metal ;) haha | 15:52 |
mgariepy | it needs some love | 15:52 |
*** rpittau is now known as rpittau|afk | 15:54 | |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_cloudkitty master: Reduce number of processes on small systems https://review.opendev.org/760617 | 16:05 |
jamesdenton | noonedeadpunk Do you happen to know if this was intentionally left master vs stable/ussuri? https://github.com/openstack/openstack-ansible/blob/stable/ussuri/playbooks/defaults/repo_packages/openstack_services.yml#L222-L226 | 16:07 |
noonedeadpunk | jamesdenton: it wasn't forked these days.... | 16:08 |
noonedeadpunk | I think it should be patched | 16:08 |
jamesdenton | oh really? | 16:08 |
noonedeadpunk | not 100%, maybe jsut blindly backported | 16:09 |
miloa | noonedeapunk: after applying the modification on glance, I ve got this error : FAILED! => {"msg": "Unexpected templating type error occurred on ({{ openstack_service_venv_bin | dirname }}/lib/): expected str, bytes or os.PathLike object, not bool"} | 16:09 |
noonedeadpunk | jamesdenton: yeah it was backport https://review.opendev.org/#/c/747108/ | 16:09 |
noonedeadpunk | my fault:( | 16:10 |
miloa | in task : TASK [ceph_client : Find the venv's python version] | 16:10 |
jamesdenton | kk no worries. | 16:10 |
jamesdenton | we may want to fix before people upgrade, trying to go back is resulting in some DB migration issues | 16:11 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_gnocchi master: Reduce number of processes on small systems https://review.opendev.org/760624 | 16:12 |
noonedeadpunk | yeah.. .and write reno maybe... | 16:13 |
jamesdenton | definitely | 16:13 |
noonedeadpunk | well, considering latest change to migrations was 1 week ago and we did bump 3 weeks ago - no db changes were made from U | 16:14 |
*** yann-kaelig has joined #openstack-ansible | 16:14 | |
noonedeadpunk | uh, no, wrong again | 16:15 |
jamesdenton | well, my deploy threw an error but i'm not seeing any ironic-inspector migrations/db changes in last 7 months, so it may be OK just need to work through it | 16:17 |
jamesdenton | well, i take that back. two in last 4 days. lol but before that, 7 mos | 16:17 |
noonedeadpunk | or not.... | 16:17 |
noonedeadpunk | https://opendev.org/openstack/ironic-inspector/commits/branch/master/ironic_inspector/migrations/versions | 16:18 |
noonedeadpunk | it seems that e2c8f9fd7b is included in Ussuri | 16:18 |
jamesdenton | https://pastebin.com/WEMjGCGS | 16:18 |
jamesdenton | b55109d5063a is what it was complaining about here | 16:19 |
noonedeadpunk | um | 16:22 |
noonedeadpunk | tbh.... | 16:22 |
noonedeadpunk | it feels like this migration is generally broken | 16:22 |
jamesdenton | well, i could probably just drop the ironic_inspector db and be OK :D | 16:23 |
jamesdenton | it's not like i was using it | 16:23 |
noonedeadpunk | uh, whatever, I got mixed up in how alembic works... | 16:24 |
jamesdenton | i just ran delete from alembic_version where version_num='b55109d5063a';, let's see what haoppens | 16:25 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_heat master: Reduce number of processes on small systems https://review.opendev.org/760630 | 16:27 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_ironic master: Reduce number of processes on small systems https://review.opendev.org/760632 | 16:30 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_magnum master: Reduce number of processes on small systems https://review.opendev.org/760633 | 16:32 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_mistral master: Reduce number of processes on small systems https://review.opendev.org/760636 | 16:35 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_manila master: Reduce number of processes on small systems https://review.opendev.org/760637 | 16:36 |
jamesdenton | noonedeadpunk so, it turns out that b55109d5063a_added_scope_column_to_rules_table.py was missing from the migrations directory. *shrug* | 16:41 |
miloa | noonedeadpunk: I emptied the ansible cache and try again, but still have this error : "Unexpected templating type error occurred on ({{ openstack_service_venv_bin | dirname }}/lib/): expected str, bytes or os.PathLike object, not bool"}. It looks like the when statement in ceph_client/taskc/main.yml is not working or perhaps i miss something. | 16:45 |
spatel | noonedeadpunk: https://docs.openstack.org/project-deploy-guide/openstack-ansible/latest/deploymenthost.html very curious about why we need to install "dnf group install "Development Tools" ? | 16:48 |
spatel | may be for pip venv etc.. | 16:49 |
openstackgerrit | Merged openstack/openstack-ansible-os_cinder master: Reduce number of processes on small systems https://review.opendev.org/759916 | 17:12 |
miloa | noonedeadpunk: finally it works, my bad, a typo when i add the patch in our dev. Many Thanks :) | 17:35 |
*** MickyMan77 has joined #openstack-ansible | 17:37 | |
*** MickyMan77 has quit IRC | 17:46 | |
*** yann-kaelig has quit IRC | 18:00 | |
*** ts103706 has joined #openstack-ansible | 18:02 | |
openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible-os_neutron master: [WIP] adjust some ovn tasks https://review.opendev.org/760647 | 18:19 |
openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible master: [DNM] split the gateway service. https://review.opendev.org/760384 | 18:22 |
*** lkoranda has quit IRC | 18:32 | |
*** miloa has quit IRC | 18:34 | |
ts103706 | Does anyone have any examples of configuring bridges for Centos8? We're able to create instances, but they are unreachable. | 18:44 |
admin0 | ts103706, do a virsh dumpxml and check if they are correctly mapped | 19:02 |
*** mike44333 has joined #openstack-ansible | 19:26 | |
ts103706 | @admin0: the source bridge and target device from virsh dumpxml both show up as connection in network manager. | 19:30 |
spatel | noonedeadpunk: deploying new osa and when i am running openstack-ansible os-keystone-install.yml ( during one setup its trying to talk to http://10.64.0.101:5000, instead of https://) | 19:50 |
spatel | may be i have internal_vip and external_vip on same IP.. | 19:51 |
spatel | that could be possible | 19:51 |
openstackgerrit | James Denton proposed openstack/openstack-ansible-os_ironic master: Fix inspector listener on bare metal https://review.opendev.org/760660 | 20:11 |
*** cshen has quit IRC | 20:16 | |
*** yolanda has quit IRC | 20:24 | |
*** yolanda has joined #openstack-ansible | 20:25 | |
*** cshen has joined #openstack-ansible | 20:30 | |
openstackgerrit | James Denton proposed openstack/openstack-ansible-os_ironic master: Update Inspector listener address and network https://review.opendev.org/760660 | 20:32 |
*** MickyMan77 has joined #openstack-ansible | 20:51 | |
*** MickyMan77 has quit IRC | 20:59 | |
*** ts103706 has quit IRC | 21:01 | |
*** jamesdenton has quit IRC | 21:10 | |
*** recyclehero has quit IRC | 21:25 | |
*** recyclehero has joined #openstack-ansible | 21:32 | |
*** MickyMan77 has joined #openstack-ansible | 21:32 | |
*** rfolco is now known as rfolco|bbl | 22:21 | |
*** dasp_ has joined #openstack-ansible | 22:26 | |
*** dave-mccowan has quit IRC | 22:26 | |
*** MickyMan77 has quit IRC | 22:28 | |
*** dasp has quit IRC | 22:28 | |
*** cshen has quit IRC | 22:48 | |
*** cshen has joined #openstack-ansible | 22:49 | |
*** spatel has quit IRC | 23:09 | |
*** macz_ has quit IRC | 23:14 | |
*** yolanda has quit IRC | 23:33 | |
*** jamesdenton has joined #openstack-ansible | 23:39 | |
*** jamesdenton has quit IRC | 23:40 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!