| *** gkadam has joined #openstack-kayobe | 03:48 | |
| *** dougsz has joined #openstack-kayobe | 09:15 | |
| mgoddard | morning all | 09:18 |
|---|---|---|
| dougsz | morning | 09:18 |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Support including and excluding files from config save https://review.openstack.org/629425 | 09:21 |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Update development scripts for control plane deployment https://review.openstack.org/634274 | 09:24 |
| *** gkadam_ has joined #openstack-kayobe | 09:55 | |
| *** gkadam has quit IRC | 09:58 | |
| mgoddard | rocky release list looking quite good now | 10:00 |
| mgoddard | if anyone has any topics for this afternoons meeting, please add them to https://wiki.openstack.org/wiki/Meetings/Kayobe | 10:00 |
| *** gkadam__ has joined #openstack-kayobe | 10:02 | |
| *** gkadam_ has quit IRC | 10:05 | |
| openstackgerrit | Doug Szumski proposed openstack/kayobe master: Add commands to run command on hosts https://review.openstack.org/589112 | 11:09 |
| *** verdurin has joined #openstack-kayobe | 11:23 | |
| openstackgerrit | Merged openstack/kayobe master: use include_tasks and import_playbook instead of include https://review.openstack.org/612563 | 13:33 |
| openstackgerrit | Doug Szumski proposed openstack/kayobe master: Remove support for deploying Prometheus node-exporter https://review.openstack.org/634689 | 13:39 |
| openstackgerrit | Doug Szumski proposed openstack/kayobe master: Remove support for deploying cAdvisor https://review.openstack.org/634690 | 13:47 |
| *** priteau has joined #openstack-kayobe | 13:53 | |
| *** oneswig has joined #openstack-kayobe | 13:57 | |
| brtknr | mgoddard: yo! | 13:59 |
| mgoddard | brtknr: hey | 13:59 |
| mgoddard | #startmeeting kayobe | 14:00 |
| openstack | Meeting started Mon Feb 4 14:00:19 2019 UTC and is due to finish in 60 minutes. The chair is mgoddard. Information about MeetBot at http://wiki.debian.org/MeetBot. | 14:00 |
| openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 14:00 |
| *** openstack changes topic to " (Meeting topic: kayobe)" | 14:00 | |
| openstack | The meeting name has been set to 'kayobe' | 14:00 |
| mgoddard | Hi folks | 14:00 |
| mgoddard | #topic rollcall | 14:00 |
| oneswig | o/ | 14:00 |
| *** openstack changes topic to "rollcall (Meeting topic: kayobe)" | 14:00 | |
| jovial[m] | o/ | 14:00 |
| mgoddard | \o | 14:00 |
| dougsz | o| | 14:01 |
| priteau | o/ | 14:01 |
| oneswig | ô | 14:01 |
| brtknr | o/ | 14:01 |
| mgoddard | #topic agenda | 14:03 |
| *** openstack changes topic to "agenda (Meeting topic: kayobe)" | 14:03 | |
| mgoddard | * Roll-call | 14:03 |
| mgoddard | * Announcements | 14:03 |
| mgoddard | ** priteau joins kayobe-core! | 14:03 |
| mgoddard | * Discussion | 14:03 |
| mgoddard | ** Rocky release status https://etherpad.openstack.org/p/kayobe-rocky-release | 14:03 |
| mgoddard | ** Stein release planning https://etherpad.openstack.org/p/kayobe-stein-release | 14:03 |
| mgoddard | * Open Discussion | 14:03 |
| mgoddard | #topic announcements | 14:03 |
| *** openstack changes topic to "announcements (Meeting topic: kayobe)" | 14:03 | |
| mgoddard | #info priteau joins kayobe-core! | 14:03 |
| mgoddard | Congrats priteau, welcome to the team | 14:04 |
| oneswig | +1 | 14:04 |
| mgoddard | Any other announcements? | 14:04 |
| priteau | Thanks, will do my best to improve Kayobe :-) | 14:04 |
| brtknr | mgoddard: you missed my last minute agenda insertion :P | 14:04 |
| dougsz | welcome aboard priteau | 14:04 |
| brtknr | +1 to priteau | 14:05 |
| mgoddard | oops sorry brtknr. We'll cover it later | 14:05 |
| mgoddard | #topic discussion | 14:05 |
| *** openstack changes topic to "discussion (Meeting topic: kayobe)" | 14:05 | |
| mgoddard | #link https://etherpad.openstack.org/p/kayobe-rocky-release | 14:05 |
| mgoddard | ^ Rocky release status | 14:05 |
| mgoddard | looking pretty good, all features are in | 14:06 |
| priteau | Etherpad not loading for me :( | 14:06 |
| mgoddard | a few bug fixes left | 14:06 |
| dougsz | I just pushed a couple of changes to remove Prometheus node-exporter and cAdvisor now that Kolla-Ansible can deploy them (in Rocky). Shall I add those to the list? | 14:07 |
| mgoddard | dougsz: could do. not too large? | 14:07 |
| dougsz | Not large, they just delete overcloud post deploy playbooks and a couple of roles | 14:07 |
| mgoddard | I'd like to make a release this week, if possible, or early next week | 14:08 |
| mgoddard | add them to the list, I expect we can manage it | 14:08 |
| dougsz | will do, thanks | 14:08 |
| mgoddard | jovial[m]: were you planning to fix up the serial consoles patch before rocky? | 14:09 |
| mgoddard | needs at least a merge fix | 14:09 |
| jovial[m] | mgoddard: yep, I'll try and fix that up shortly | 14:10 |
| mgoddard | jovial[m]: great | 14:10 |
| mgoddard | anything else we need to be doing re Rocky? | 14:10 |
| dougsz | nothing else in the pipeline from me | 14:11 |
| priteau | I tried to reproduce my issue with IP allocation but didn't manage, I will give it another try tomorrow morning. | 14:11 |
| mgoddard | priteau: is that with or without the network-allocations.yml fix? | 14:12 |
| priteau | I also still have to push Cumulus patches, but don't hold the release just for me. | 14:12 |
| priteau | mgoddard: I think I saw the issue even with an existing network-allocations.yml file existing, but couldn't reproduce yet. Maybe I misremembered. | 14:13 |
| mgoddard | ok, keep an eye out for it | 14:13 |
| mgoddard | let's move onto Stein | 14:13 |
| mgoddard | #link https://etherpad.openstack.org/p/kayobe-stein-release | 14:14 |
| mgoddard | I think etherpad and gerrit are also not working for me | 14:14 |
| mgoddard | there's not too much on the stein etherpad yet | 14:14 |
| mgoddard | if there are features you'd like to see in stein, please add them to that page. Perhaps we need a new 'feature requests' section | 14:15 |
| mgoddard | or 'provisional feature list' | 14:15 |
| mgoddard | or something | 14:15 |
| oneswig | Would that be a good point to integrate better support for Ceph-Ansible? | 14:15 |
| mgoddard | or priorities | 14:16 |
| jovial[m] | yeah, the etherpad won't load for me either | 14:16 |
| mgoddard | oneswig: yes I think so | 14:16 |
| oneswig | OK it's added | 14:17 |
| dougsz | I added another for better support for out-of-band provisioned overcloud nodes | 14:17 |
| mgoddard | that's a good one | 14:17 |
| *** gkadam__ has quit IRC | 14:18 | |
| priteau | What about importing an existing Kolla / K-A installation? | 14:18 |
| dougsz | +1, that was a nice one | 14:19 |
| mgoddard | priteau: that would be a good one | 14:19 |
| mgoddard | etherpad seems ok again now | 14:20 |
| mgoddard | ok, let's move on | 14:23 |
| mgoddard | #topic Deploying magnum in all-in-one VM | 14:23 |
| *** openstack changes topic to "Deploying magnum in all-in-one VM (Meeting topic: kayobe)" | 14:23 | |
| mgoddard | brtknr: take it away | 14:23 |
| brtknr | oops | 14:25 |
| brtknr | I am in the process of deploying magnum on an existing all-in-one kayobe deployment | 14:26 |
| brtknr | This is what ive done so far: | 14:26 |
| brtknr | vi config/src/kayobe-config/etc/kolla/globals.yml | 14:26 |
| brtknr | # enabled_magnum: True | 14:26 |
| brtknr | kayobe overcloud container image build magnum --push | 14:26 |
| brtknr | This is what I intend to do next: | 14:27 |
| brtknr | kayobe overcloud container image pull --kolla-tags magnum --kolla-skip-tags common | 14:27 |
| brtknr | kayobe overcloud service upgrade --kolla-tags magnum --kolla-skip-tags commony | 14:27 |
| brtknr | I have a feeling upgrate is not the right instruction to use if there is no existing magnum deployment | 14:27 |
| priteau | You should be able to do a `service deploy` | 14:28 |
| mgoddard | brtknr: that's right - the initial deploy needs to use 'deploy', as it causes the DBs to be created | 14:28 |
| priteau | Does it make sense to use both kolla-tags and kolla-skip-tags? | 14:29 |
| brtknr | so `kayobe overcloud service deploy --kolla-tags magnum --kolla-skip-tags common`? | 14:29 |
| mgoddard | you shouldn't need to both build and pull the image. If the one on dockerhub is fine, a pull should suffice | 14:29 |
| mgoddard | that should do the trick | 14:29 |
| dougsz | Yeah, you'll want the common role to run to configure fluentd logging, so It'd be tempted not to skip that one | 14:29 |
| dougsz | (unless it's already deployed on the node running magnum) | 14:30 |
| priteau | IIRC there are also important tags, like config, which you will need | 14:30 |
| priteau | When in doubt, don't use tags! | 14:30 |
| dougsz | And also, is there any Magnum HAProxy config? If so you'll want to run that role too.. | 14:30 |
| oneswig | horizon too? | 14:31 |
| brtknr | I intend to also run kayobe overcloud service reconfigure --kolla-tags haproxy | 14:31 |
| mgoddard | +1 for horizon and haproxy | 14:31 |
| brtknr | Will kayobe overcloud service reconfigure --kolla-tags haproxy horizon do the job? | 14:32 |
| mgoddard | kayobe overcloud service reconfigure --kolla-tags haproxy,horizon | 14:33 |
| mgoddard | that should work | 14:33 |
| brtknr | Cool, I will report separately on progress | 14:33 |
| mgoddard | as priteau says though, sometimes tags add more time if you get it wrong and need to do it again, sometimes just a deploy with no tags works best | 14:34 |
| brtknr | Image build failed on my all-in-one VM but pull appears to have worked | 14:34 |
| mgoddard | ok, ask on here if you have issues brtknr | 14:34 |
| mgoddard | let's move on | 14:34 |
| mgoddard | #topic Open discussion | 14:35 |
| *** openstack changes topic to "Open discussion (Meeting topic: kayobe)" | 14:35 | |
| mgoddard | anyone have anything else to discuss? | 14:35 |
| brtknr | mgoddard: thanks :) | 14:35 |
| mgoddard | sounds like no, everyone gets 24 minutes back | 14:36 |
| mgoddard | thanks everyone, have a nice day | 14:36 |
| mgoddard | #endmeeting | 14:36 |
| priteau | I have something | 14:36 |
| *** openstack changes topic to "Deployment of containerised OpenStack to bare metal | https://wiki.openstack.org/wiki/Meetings/Kayobe" | 14:36 | |
| openstack | Meeting ended Mon Feb 4 14:36:50 2019 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 14:36 |
| openstack | Minutes: http://eavesdrop.openstack.org/meetings/kayobe/2019/kayobe.2019-02-04-14.00.html | 14:36 |
| openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/kayobe/2019/kayobe.2019-02-04-14.00.txt | 14:36 |
| openstack | Log: http://eavesdrop.openstack.org/meetings/kayobe/2019/kayobe.2019-02-04-14.00.log.html | 14:36 |
| mgoddard | #undo | 14:36 |
| priteau | Oops, too late | 14:37 |
| mgoddard | sorry priteau! | 14:37 |
| priteau | I don't think undo works if the meeting has already ended | 14:37 |
| priteau | Sorry, I was looking for a link | 14:37 |
| priteau | I wanted to ask if it was possible to get rid of https://docs.openstack.org/releasenotes/kayobe | 14:37 |
| priteau | It might be confusing if people found them, as they are not updated anymore. | 14:37 |
| mgoddard | good point, thanks | 14:37 |
| mgoddard | I'll ask in #openstack-infra | 14:38 |
| priteau | Also it looks like https://kayobe-release-notes.readthedocs.io/en/latest/unreleased.html doesn't have the latest merges | 14:38 |
| priteau | I would expect to see "Adds support for specifying a CA certificate when accessing APIs" | 14:38 |
| mgoddard | priteau: that's right, I couldn't find a way to trigger two RTD builds in zuul | 14:40 |
| mgoddard | it's not actually released yet, so probably not an issue | 14:40 |
| mgoddard | but we do need to remember to trigger a build | 14:40 |
| mgoddard | perhaps I could add a periodic build | 14:40 |
| mgoddard | I don't think it's possible | 14:42 |
| priteau | I confirm that the network-allocation.yml patch does *not* fix the issue. I was able to reproduce it using a fresh Kayobe install in a VM and just trying to configure a seed host. | 15:23 |
| priteau | 1) Usual Kayobe setup and control host bootstrap | 15:23 |
| priteau | 2) Copy network-allocation.yml from https://review.openstack.org/#/c/631557/ to $KAYOBE_CONFIG_PATH | 15:24 |
| priteau | 3) Configure an admin network and set it to be the network attached to the seed | 15:25 |
| priteau | 4) running `kayobe seed host configure` reports: | 15:25 |
| priteau | TASK [ip-allocation : Ensure IP addresses are allocated] changed: [testseed -> localhost] => (item={'cidr': u'192.168.100.0/24', 'allocation_pool_start': u'192.168.100.1', 'net_name': u'admin_oc_net', 'allocation_pool_end': u'192.168.100.10'}) => {"changed": true, "ip": "192.168.100.1", "item": {"allocation_pool_end": "192.168.100.10", "allocation_pool_start": "192.168.100.1", "cidr": "192.168.100.0/24", "net_name": "admin_oc_net"}} | 15:26 |
| priteau | So IP allocation works fine | 15:26 |
| priteau | But then later on: | 15:26 |
| priteau | TASK [MichaelRigart.interfaces : Create the network configuration file for ethernet devices] | 15:26 |
| priteau | changed: [testseed] => (item={'device': u'eth0:0', 'netmask': u'255.255.255.0', 'bootproto': u'static', 'onboot': u'yes', 'address': u'0.0.0.0'}) => {"changed": true, "checksum": "92992128a4e8cae492696a68f1d12ea8593c9476", "dest": "/etc/sysconfig/network-scripts/ifcfg-eth0:0", "gid": 0, "group": "root", "item": {"address": "0.0.0.0", "bootproto": "static", "device": "eth0:0", "netmask": "255.255.255.0", "onboot": "yes"}, "md5sum": " | 15:26 |
| priteau | cecc70fd41acb221aad5ee648b017e4b", "mode": "0644", "owner": "root", "size": 101, "src": "/home/stack/.ansible/tmp/ansible-tmp-1549293965.94-87692785404031/source", "state": "file", "uid": 0} | 15:26 |
| priteau | Note the 0.0.0.0 address | 15:26 |
| priteau | Kayobe subsequently fails on: RUNNING HANDLER [MichaelRigart.interfaces : Check active Ethernet interface state] | 15:27 |
| priteau | IP allocation and network configuration is done via a single invocation of ansible-playbook, so I am thinking it is caused by vars not being updated? | 15:29 |
| brtknr | mgoddard, priteau: after running all those commands, kayobe reports that no changes were made, and indeed there is no endpoint for magnum when i run `openstack coe cluster list` | 15:57 |
| priteau | brtknr: Did you try running without tags? | 15:58 |
| mgoddard | priteau: what's with the eth0:0? | 15:58 |
| brtknr | priteau: i'll do that now | 15:58 |
| priteau | mgoddard: That's the interface I specified for this network. I didn't want to lose SSH access. | 15:59 |
| mgoddard | priteau: and after a subsequent run it works? | 16:00 |
| mgoddard | priteau: could you check the content of network-allocation.yml | 16:02 |
| priteau | mgoddard: Yes, subsequent run works. | 16:05 |
| mgoddard | priteau: hmm, perhaps it's not reloading network-allocation.yml after ip-allocation.yml runs | 16:06 |
| mgoddard | priteau: we probably need to add a separate call to ansible-playbook | 16:06 |
| priteau | This is what network-allocation.yml looks like after the run that allocates the IP | 16:07 |
| priteau | admin_oc_net_ips: | 16:07 |
| priteau | testseed: 192.168.100.1 | 16:07 |
| mgoddard | priteau: something like this? http://paste.openstack.org/show/744506/ | 16:08 |
| priteau | Looks like it would do the right thing, I will test it | 16:10 |
| mgoddard | priteau: thanks | 16:12 |
| priteau | ip-allocation already does a set_fact so I am not sure why the current code doesn't work as expected | 16:12 |
| priteau | It seems to be related either to new networks or new hosts, as I wasn't able to reproduce when simply adding an existing network to an existing host | 16:13 |
| openstackgerrit | Will Szumski proposed openstack/kayobe master: Add commands to run command on hosts https://review.openstack.org/589112 | 16:15 |
| priteau | mgoddard: your patch worked! I assume the same approach needs to be added for seed-hypervisor and overcloud | 16:38 |
| mgoddard | priteau: great :) yes it'll need to be applied to those places | 16:41 |
| *** priteau has quit IRC | 16:53 | |
| brtknr | priteau: So after I ran `kayobe overcloud service deploy` without any tags, there were no magnum related chages reported | 16:54 |
| brtknr | wait, somehow things got reset back to enable_magnum: False | 16:56 |
| brtknr | Or I forgot to save globals.yml? | 16:56 |
| * jovial[m] sent a long message: < https://matrix.org/_matrix/media/v1/download/matrix.org/yyMwbDRLJEOPWBrsCUMyMnXR > | 16:57 | |
| dougsz | brtknr, maybe you made the change in etc/kolla/globals.yml | 16:57 |
| brtknr | dougsz: Am I not supposed to? | 16:58 |
| dougsz | ah, what jovial said. | 16:58 |
| jovial[m] | no, that file gets generated by kayobe | 16:58 |
| dougsz | That will get overwritten by Kayobe - it's the Kolla Ansible config that is generated by Kayobe | 16:59 |
| jovial[m] | so will be overwritten when you reun kayobe | 16:59 |
| brtknr | Oh i see | 16:59 |
| dougsz | Never seen that matrix.org thing jovial! | 16:59 |
| brtknr | So I ought to set kolla_enable_magnum: True inside `config/src/kayobe-config/etc/kayobe/kolla.yml` | 17:00 |
| brtknr | So I ought to set `kolla_enable_magnum: True` inside `config/src/kayobe-config/etc/kayobe/kolla.yml` | 17:01 |
| dougsz | that sounds more like it | 17:01 |
| *** priteau has joined #openstack-kayobe | 17:05 | |
| brtknr | Do I need to explictly pull an image or would `kayobe overcloud service deploy` also handle the pulling by default | 17:07 |
| jovial[m] | you need to explicitly pull | 17:08 |
| priteau | brtknr: It would pull automatically. | 17:08 |
| priteau | jovial[m]: Seems like we disagree... | 17:08 |
| jovial[m] | I may well be wrong :D | 17:08 |
| brtknr | lol... | 17:08 |
| brtknr | well, i am currently running overcloud service deploy without pulling explictly so lets see if it runs to the end | 17:09 |
| mgoddard | it will pull automatically. Sometimes you need to pull when doing an upgrade if there is a new image with the same tag | 17:13 |
| jovial[m] | ahh, thats what I was remembering | 17:13 |
| mgoddard | jovial[m]: base64! | 17:14 |
| mgoddard | jovial[m]: nice idea | 17:15 |
| jovial[m] | mgoddard: I was wondering what code review would make of that - couldn't find a nicer way though :D | 17:16 |
| mgoddard | jovial[m]: if it works... | 17:17 |
| jovial[m] | its a shame you can't use !unsafe with extra-vars passed in via the command line | 17:17 |
| jovial[m] | https://docs.ansible.com/ansible/latest/user_guide/playbooks_advanced_syntax.html | 17:17 |
| mgoddard | jovial[m]: I think you can pass in JSON | 17:18 |
| mgoddard | not sure about YAML | 17:18 |
| jovial[m] | it looks like you might be able to pass yaml aswell (looking at https://docs.ansible.com/ansible/latest/user_guide/playbooks_variables.html) | 17:21 |
| *** oneswig has quit IRC | 17:22 | |
| *** dougsz has quit IRC | 17:57 | |
| openstackgerrit | Will Szumski proposed openstack/kayobe master: Automatically enable serial consoles on post configure https://review.openstack.org/614164 | 18:07 |
| openstackgerrit | Mark Goddard proposed openstack/kayobe master: Support configuration of yum.conf https://review.openstack.org/634788 | 19:45 |
| brtknr | Whats the purpose of running `kayobe overcloud service reconfigure --kolla-tags haproxy,horizon`? I am able to run `openstack coe cluster *` commands without running this. | 21:30 |
| brtknr | Does running `kayobe overcloud service deploy` somehow also remove the need for reconfigure? | 21:31 |
| brtknr | Is service reconfigure only required when you upgrade? | 21:31 |
| brtknr | Or in cases where it not a all-in-one deployment | 21:31 |
| brtknr | Or in cases where it not an all-in-one deployment? | 21:31 |
| *** priteau has quit IRC | 22:20 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!