| *** luksky has quit IRC | 00:09 | |
| *** DanyC has quit IRC | 00:24 | |
| kplant | 23/quit | 00:37 |
|---|---|---|
| *** kplant has quit IRC | 00:37 | |
| jamesdenton | djhankb You can use this as a reference: https://docs.openstack.org/openstack-ansible-os_neutron/latest/app-openvswitch.html#openstack-ansible-user-variables | 00:57 |
| jamesdenton | The scenario here is for OVS, but it would apply to LXB as well. You would basically define the 'neutron_provider_networks' with your desired bridge mappings and interface | 00:58 |
| djhankb | jamesdenton: Thank you, I was able to find that document and get it partially configured. | 00:58 |
| jamesdenton | Is there something in particular you're hoping to achieve? | 00:59 |
| djhankb | in my lab deployment here, I was able to get Neutron running well by defining both physical_interface_mappings as well as bridge_mappings under linux_bridge. | 00:59 |
| djhankb | I used the aforementioned documentation and am now getting Ansible to configure physical_interface_mappings. the "bridge_mappings" part may or may not be needed, I am still learning how all of this stuff works! | 01:00 |
| djhankb | I had recently added my own SSL certificates to HAProxy as well as rabbitmq - and upon re-running playbooks I broke my neutron l3 agent. Adding "physical_interface_mappings" via neutron_provider_networks in user_variables.yml and I was still having issues until I restarted rabbitmq (stop_app / start_app) and I think everything is working... | 01:05 |
| jamesdenton | i believe bridge_mappings is meant for OVS, while physical_interface_mappings is for LXB (as LXB will create a bridge for each network and tag off the mapped interface). Whereas with OVS, you're mapping the provider name to an OVS bridge, like br-ex or br-provider. Not sure i trust the config file example TBH | 01:06 |
| djhankb | Right, right. Either way I was ultimately trying to find out how to add "bridge_mappings" via user_variables.yml, and I *think* I can interpret this documentation as yml: network_mappings == [linux_bridge]:physical_interface_mappings. I can only assume that yml:network_interface_mappings == [linux_bridge]:bridge_mappings... | 01:11 |
| djhankb | I'm just trying to do *everything* using Ansible as I am building out this lab so that I can keep my config in Git and reproduce this once I am comfortable enough to build Openstack in Production | 01:12 |
| jamesdenton | Well, if you define the 'provider' network in openstack_user_config.yml, then most everything should be handled for you. no need to specify overrides at all | 01:29 |
| jamesdenton | So for example: https://github.com/openstack/openstack-ansible/blob/master/etc/openstack_deploy/openstack_user_config.yml.example#L278-L286 | 01:30 |
| jamesdenton | actually, let me enhance that. one sec | 01:31 |
| jamesdenton | https://pasted.tech/pastes/655df585faaf48610db58372cbfb9fc1f8ed365d | 01:32 |
| jamesdenton | What you'll end up with is physical_interface_mapping = physnet1:eth1 | 01:32 |
| jamesdenton | which means any time you create a neutron provider network using 'physnet1' label, the agent will use eth1.X in the respective bridges it creates. | 01:33 |
| jamesdenton | for this conversation, you can ignore container_interface. Leave it as-is | 01:33 |
| jamesdenton | and for the 'range', you would want to use the VLANs you've allocated to tenant networks (if any) | 01:34 |
| djhankb | That is a good point, I do have provider networks defined, and this one I remember adding later, and manually... | 01:42 |
| djhankb | It looks like the piece i was missing was the "host_bind_override" you pointed out | 01:44 |
| djhankb | in mine I had the same value for both container interface as well as host_bind_override... so that is likely my problem, which I am trying to fix the wrong way. | 01:45 |
| *** djhankb has quit IRC | 03:56 | |
| *** ansmith_ has joined #openstack-ansible | 04:39 | |
| *** ansmith has quit IRC | 04:40 | |
| *** eglute has quit IRC | 07:15 | |
| *** hamzaachi has joined #openstack-ansible | 07:16 | |
| openstackgerrit | Dmitriy Rabotjagov (noonedeadpunk) proposed openstack/openstack-ansible-lxc_hosts stable/rocky: Fix image fetching process for leap 15 https://review.openstack.org/648854 | 07:45 |
| *** DanyC has joined #openstack-ansible | 08:15 | |
| *** DanyC has quit IRC | 08:29 | |
| *** cshen has joined #openstack-ansible | 08:51 | |
| *** hamzaachi_ has joined #openstack-ansible | 09:05 | |
| *** cshen has quit IRC | 09:08 | |
| *** hamzaachi has quit IRC | 09:08 | |
| *** hamzaachi__ has joined #openstack-ansible | 09:10 | |
| *** hamzaachi_ has quit IRC | 09:13 | |
| *** hamzaachi__ has quit IRC | 09:17 | |
| *** hamzaachi has joined #openstack-ansible | 09:17 | |
| *** hamzaachi_ has joined #openstack-ansible | 09:19 | |
| *** hamzaachi has quit IRC | 09:22 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_horizon master: Stop installing extra packages for Horizon https://review.openstack.org/648842 | 09:45 |
| *** cshen has joined #openstack-ansible | 11:04 | |
| *** luksky has joined #openstack-ansible | 11:08 | |
| *** cshen has quit IRC | 11:08 | |
| *** ianychoi has quit IRC | 11:36 | |
| *** cshen has joined #openstack-ansible | 13:04 | |
| *** cshen has quit IRC | 13:08 | |
| *** shyamb has joined #openstack-ansible | 15:01 | |
| *** cshen has joined #openstack-ansible | 15:01 | |
| *** shyamb has quit IRC | 15:13 | |
| *** cshen has quit IRC | 15:33 | |
| openstackgerrit | Merged openstack/ansible-role-python_venv_build master: Revert "speedup: move when block to outside include" https://review.openstack.org/648840 | 16:25 |
| mnaser | hmm, looks like I just caught a bug in python_venv_build | 16:34 |
| mnaser | odyssey4me: it seems that it's possible that the container python_venv_build uses to build it's wheels *is not* the same as the one running lsyncd, which results in a 404. I have no idea how to end in this place, because it seems to be happening undeterministically, but I also haven't had much luck understanding the logic of selection | 16:35 |
| mnaser | and how a wheel builder host is picked | 16:35 |
| mnaser | I guess we don't see this in CI because there is only repo_server container | 16:36 |
| mnaser | ceph_client is oddly broken on centos. im checking.. https://www.irccloud.com/pastebin/lg5P3YRn/ | 16:56 |
| mnaser | that's the "ceph_client : Add ceph repo" task | 16:56 |
| mnaser | aw crap | 16:57 |
| mnaser | I think its the `ceph_gpg_keys` reference | 16:57 |
| mnaser | yup it is | 16:57 |
| *** hamzaachi_ has quit IRC | 17:00 | |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-ceph_client master: Remove unnecessary GPG keys setting for YUM repos. https://review.openstack.org/648877 | 17:01 |
| mnaser | cores: please have a look at https://review.openstack.org/#/q/topic:stein-deploy -- ill continue to push things that I see are upgrade breaking / regressions | 17:02 |
| *** hamzaachi_ has joined #openstack-ansible | 17:03 | |
| *** cshen has joined #openstack-ansible | 17:30 | |
| *** cshen has quit IRC | 17:34 | |
| *** hamzaachi_ has quit IRC | 17:42 | |
| *** cshen has joined #openstack-ansible | 17:45 | |
| *** cshen has quit IRC | 17:49 | |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible master: glance: stop using healthcheck endpoint https://review.openstack.org/648878 | 17:50 |
| *** DanyC has joined #openstack-ansible | 17:59 | |
| *** DanyC has quit IRC | 18:17 | |
| *** pcaruana has joined #openstack-ansible | 19:32 | |
| *** tosky has joined #openstack-ansible | 19:34 | |
| *** pcaruana has quit IRC | 19:40 | |
| *** cshen has joined #openstack-ansible | 19:45 | |
| *** cshen has quit IRC | 19:49 | |
| *** ArchiFleKs has joined #openstack-ansible | 20:21 | |
| *** ivve has joined #openstack-ansible | 21:11 | |
| *** DanyC has joined #openstack-ansible | 21:16 | |
| *** cshen has joined #openstack-ansible | 21:45 | |
| *** cshen has quit IRC | 21:49 | |
| *** tosky has quit IRC | 22:06 | |
| *** DanyC has quit IRC | 22:24 | |
| *** ianychoi has joined #openstack-ansible | 22:28 | |
| *** luksky has quit IRC | 22:34 | |
| openstackgerrit | Merged openstack/openstack-ansible master: glance: stop using healthcheck endpoint https://review.openstack.org/648878 | 22:47 |
| *** ivve has quit IRC | 22:54 | |
| *** nurdie has joined #openstack-ansible | 22:57 | |
| nurdie | I don't have a dedicated network host in my OSA environment. In Pike, neutron-agents were running in containers on 3 controllers. After the Queens upgrade, they are gone and and I have no neutron agents | 22:58 |
| nurdie | What did I miss? o_0 | 22:58 |
| nurdie | That's a lie. I have neutron agents. I have no neutron-dhcp-agents though | 23:00 |
| nurdie | How do I get OSA to install dhcp agents? | 23:12 |
| guilhermesp | nurdie since queens neutron-agents runs on baremetal. | 23:38 |
| nurdie | guilhermesp: Yep, I get that, except mine isn't installing them lol | 23:39 |
| nurdie | I have no L3 agents, no DHCP agents, etc. The only neutron agents I have are the linuxbridge-agents running on compute nodes | 23:39 |
| guilhermesp | hum, no neutron process at all on controller nodes? How did you run the upgrade? | 23:43 |
| nurdie | run-upgrade.sh for half. We encountered a galera bug, fixed it, and proceeded using https://docs.openstack.org/openstack-ansible/queens/admin/upgrades/major-upgrades.html | 23:45 |
| nurdie | guilhermesp: When I rerun setup-hosts --limit "neutron_agent", it still doesn't install them | 23:45 |
| nurdie | I could install the venvs and systemd unit files manually, but would be cleaner to figure out why OSA isn't seeing that it should | 23:46 |
| nurdie | And faster lol | 23:46 |
| guilhermesp | all right so you ran the upgrade using run-upgrade.sh script. Failed, and then you re-ran again. Did it finish successfully ? | 23:53 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!