| *** faizy98 has quit IRC | 00:01 | |
| *** aredinger has joined #openstack-ansible | 00:54 | |
| *** aredinger has quit IRC | 00:59 | |
| *** mmercer has quit IRC | 01:05 | |
| *** spatel has joined #openstack-ansible | 01:41 | |
| *** spatel has quit IRC | 01:46 | |
| *** spatel has joined #openstack-ansible | 01:51 | |
| *** aredinger has joined #openstack-ansible | 01:54 | |
| *** aredinger has quit IRC | 01:59 | |
| *** evrardjp_ has joined #openstack-ansible | 02:12 | |
| *** evrardjp has quit IRC | 02:15 | |
| *** aredinger has joined #openstack-ansible | 02:54 | |
| *** aredinger has quit IRC | 02:59 | |
| *** spatel has quit IRC | 03:11 | |
| *** maddtux has joined #openstack-ansible | 03:21 | |
| *** chhagarw has joined #openstack-ansible | 03:42 | |
| jungleboyj | spotz: odyssey4me I figured out how to get the console to work. It is related to SSL being enabled. Need to click to open the console in its own window so you can deal with the certificate error. Then the console actually loads. | 03:51 |
|---|---|---|
| *** aredinger has joined #openstack-ansible | 03:54 | |
| *** udesale has joined #openstack-ansible | 03:56 | |
| *** aredinger has quit IRC | 03:59 | |
| *** chhavi__ has joined #openstack-ansible | 04:08 | |
| *** chhagarw has quit IRC | 04:11 | |
| *** aredinger has joined #openstack-ansible | 04:54 | |
| *** aredinger has quit IRC | 04:58 | |
| *** threestrands has quit IRC | 05:40 | |
| *** cshen has joined #openstack-ansible | 05:48 | |
| *** aredinger has joined #openstack-ansible | 05:54 | |
| *** cshen has quit IRC | 05:57 | |
| *** aredinger has quit IRC | 05:58 | |
| *** rgogunskiy has joined #openstack-ansible | 06:39 | |
| *** udesale has quit IRC | 06:49 | |
| *** aredinger has joined #openstack-ansible | 06:54 | |
| *** aredinger has quit IRC | 06:58 | |
| *** fghaas has joined #openstack-ansible | 07:01 | |
| *** cshen has joined #openstack-ansible | 07:06 | |
| *** aredinger has joined #openstack-ansible | 07:10 | |
| *** fghaas has quit IRC | 07:11 | |
| *** aredinger has quit IRC | 07:15 | |
| *** pcaruana has joined #openstack-ansible | 07:20 | |
| *** DanyC has joined #openstack-ansible | 07:48 | |
| *** DanyC has quit IRC | 07:59 | |
| *** DanyC has joined #openstack-ansible | 07:59 | |
| *** DanyC has quit IRC | 08:03 | |
| *** fghaas has joined #openstack-ansible | 08:10 | |
| *** hamzaachi has joined #openstack-ansible | 08:24 | |
| francois | good morning! | 08:53 |
| francois | I'm trying to deploy octavia with openstack-ansible queens, the network configuration is driving me crazy | 08:54 |
| francois | trying to use a vlan based provider network with the standard neutron-linuxbridge | 08:55 |
| francois | the default flat network setup mentioned in the doc looks overkill in small deployments | 08:59 |
| francois | https://docs.openstack.org/openstack-ansible-os_octavia/latest/configure-octavia.html#setup-a-neutron-network-for-use-by-octavia | 08:59 |
| francois | or maybe I'm missing something obvious... anyway would love any pointer from people that have already deployed it :) | 09:00 |
| francois | my config: https://etherpad.openstack.org/p/Q8yjASTk69 | 09:03 |
| francois | two vlan interfaces are being configured on the controller nodes br-vlan.1796 (neutron) and bond0.1796 (/etc/network/interfaces) and only bond0.1796 gets traffic | 09:07 |
| hwoarang | hello cores: anyone to get https://review.openstack.org/#/c/614355/ in? we need it for opnfv. TIA! | 09:10 |
| *** cshen has quit IRC | 09:23 | |
| *** DanyC has joined #openstack-ansible | 09:29 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_keystone stable/rocky: Add memcache flushing handler on db migrations https://review.openstack.org/613321 | 09:30 |
| hwoarang | odyssey4me: rocky + metal appears to be broken on centos and suse http://logs.openstack.org/77/614777/1/check/openstack-ansible-deploy-aio_metal-opensuse-423/1a94cb3/job-output.txt.gz#_2018-11-01_15_18_43_858094 but not sure when that regressed | 09:37 |
| hwoarang | not sure if there is an in-flight fix for that already | 09:37 |
| *** aedc has joined #openstack-ansible | 09:40 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_glance stable/rocky: zuul: Switch the Xenial distro job to Bionic https://review.openstack.org/615116 | 09:43 |
| odyssey4me | hwoarang hmm, odd that it's happening on metal only | 09:43 |
| odyssey4me | I guess it's likely because we get the host with the python env setup already with py3 I suspect... whereas with the containers we control it more | 09:43 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_nova master: zuul: Switch the Xenial lxd job to Bionic https://review.openstack.org/612959 | 09:44 |
| hwoarang | yeah looks like it. centos suse only do py2 for now | 09:44 |
| hwoarang | good catch | 09:45 |
| odyssey4me | hmm, I'm not sure I understand why the error is coming out like that though - given the instructions: http://logs.openstack.org/77/614777/1/check/openstack-ansible-deploy-aio_metal-opensuse-423/1a94cb3/logs/ara-report/result/2272ec6b-aa44-4824-b6d4-4c82f9f0488a/ | 09:47 |
| odyssey4me | oh, it must be virtualenv that's giving that error | 09:48 |
| hwoarang | it recently got regressed it seems | 09:48 |
| hwoarang | i can see builds from 31st being OK and on Nov 1st i see failures. either it regressed or there is a race somewhere | 09:49 |
| odyssey4me | the thing is that virtualenv is now trying to find python2.7 in the venv, but https://github.com/openstack/openstack-ansible-os_keystone/blob/stable/rocky/tasks/keystone_install_source.yml#L84-L90 removes it on purpose and virtualenv is supposed to put it there | 09:49 |
| odyssey4me | or something changed upstream | 09:50 |
| odyssey4me | as in the build images, or a package change for virtualenv | 09:50 |
| hwoarang | i see | 09:50 |
| hwoarang | btw any further objections on https://review.openstack.org/#/c/604366/ ? | 09:51 |
| *** hamzaachi has quit IRC | 09:51 | |
| *** aedc has quit IRC | 09:52 | |
| *** udesale has joined #openstack-ansible | 09:53 | |
| odyssey4me | hwoarang nope, lgtm | 09:53 |
| hwoarang | awesome thank you | 09:53 |
| odyssey4me | woot - we have a new option for creating images when dib releases next: https://review.openstack.org/#/c/563748/6 | 09:54 |
| odyssey4me | we should prep a similar element for suse & centos | 09:54 |
| *** evrardjp_ has quit IRC | 10:03 | |
| hwoarang | aha that's useful | 10:04 |
| openstackgerrit | Merged openstack/openstack-ansible-rabbitmq_server stable/rocky: Start rabbitmq service when config is in place https://review.openstack.org/614355 | 10:06 |
| *** cshen has joined #openstack-ansible | 10:06 | |
| *** cshen has quit IRC | 10:11 | |
| openstackgerrit | Arx Cruz proposed openstack/openstack-ansible-os_tempest master: WIP - Enable python-tempestconf support https://review.openstack.org/612968 | 10:19 |
| openstackgerrit | Damian Cikowski (dcdamien) proposed openstack/openstack-ansible-os_cinder master: Make open-iscsi as requirement just for volume and backup services https://review.openstack.org/614556 | 10:28 |
| *** udesale has quit IRC | 10:40 | |
| *** chhavi__ has quit IRC | 10:41 | |
| *** chhavi__ has joined #openstack-ansible | 10:41 | |
| *** cshen has joined #openstack-ansible | 10:50 | |
| *** electrofelix has joined #openstack-ansible | 11:00 | |
| fghaas | Quick question here before I file a bug: is running openstack-ansible with sudo considered unsupported? In rocky that is broken, at least when it's sudo without -H, because of things like this: https://github.com/openstack/openstack-ansible/blob/master/playbooks/ceph-rgw-keystone-setup.yml#L30 (openstacksdk expects to find the `clouds.yaml` file in the users $HOME, as in /home/foobar/.config, whereas openstack_openrc drops it | 11:22 |
| fghaas | in /root/config). openrc_openstack_client_config_dir_dest defaults to "{{ ansible_env.HOME }}/.config/openstack"; I'm not sure if that's perhaps an off-label use of ansible_env. | 11:22 |
| fghaas | sorry, /root/.config not /root/config. | 11:23 |
| odyssey4me | fghaas yes, we assume it is run as root - not as another user | 11:24 |
| odyssey4me | fghaas if we don't have that documented somewhere, we should do that | 11:25 |
| openstackgerrit | Merged openstack/openstack-ansible-os_nova master: Remove unnecessary package install duplication https://review.openstack.org/614333 | 11:25 |
| fghaas | odyssey4me: right. If I see "must be run as root" somewhere, then I assume it means the euid must be root, i.e. that running it with sudo is perfectly legit. | 11:25 |
| odyssey4me | there is a spec up to allow a non-root user for the deploy host, but it's not been approved and it doesn't seem to be much of a priority for anyone to actually do the work | 11:26 |
| odyssey4me | I'd personally like to see it done, but am stretched already. | 11:26 |
| fghaas | Right, like I said though, I think I'm not alone in thinking that "run this as root" only means "be sure to run the process with an euid of 0". Not "be root, and have an environment that's fully configured for root" — specifically since up to Queens just running "sudo openstack-ansible" definitely worked. | 11:27 |
| fghaas | (even without -H, that is) | 11:27 |
| odyssey4me | fghaas I'm surprised that it does, and if there's an easy fix to be done then we should do it | 11:28 |
| *** dave-mccowan has joined #openstack-ansible | 11:29 | |
| odyssey4me | for rocky we did a bunch more isolation of pip installs to keep them in venvs only, and to use the ansible modules with localhost delegation for the service setup - so that's why it's happening now and not before | 11:29 |
| odyssey4me | so it's possible that when we run the openstack_openrc role, the variables used there aren't quite right for a non-root | 11:31 |
| fghaas | Yeah I think the easy fix is to just clearly document we mandate "sudo -i" before running openstack-ansible, or instead running "sudo -H openstack-ansible". | 11:31 |
| fghaas | We could however also fail explicitly on ansible_env.HOME != /root | 11:32 |
| fghaas | Or can you run a POSIX system where root's $HOME is *not* /root? Not sure | 11:33 |
| persia | fghaas: You can (and many do). You can't run an FHS-compliant systen where /root isn't $HOME for root. | 11:34 |
| fghaas | Well yeah I meant can you run such a system without being guilty of breaking standards. :) | 11:34 |
| persia | I suspect there are enough bits of OSA that assume the base system is Linux that assuming FHS is probably safe, but if there are codepaths to support other SUS-compliant systems, then assuming FHS probably isn't safe. | 11:36 |
| fghaas | The thing with this issue is that it hits you comparatively late in the setup process, and the error message is very unintuitive for pointing you to the root cause of the problem. http://paste.openstack.org/show/734028/ I'd really like to see that being flagged a wee bit earlier. | 11:37 |
| odyssey4me | yep, agreed | 11:37 |
| odyssey4me | we should probably rather do some sort of check in the 'openstack-ansible' script to verify that it's being run as the root user, and perhaps that the right value for the home directory is present | 11:38 |
| fghaas | It already checks for the uid. It doesn't check for $HOME. | 11:38 |
| fghaas | afk for a bit, back in about an hour. But I reckon we already agree that this is a bug (either doc or functional), so I can file that later. | 11:40 |
| odyssey4me | ok, let me try and think through the original problem here - because maybe we can get it working like you'd want it to | 11:40 |
| fghaas | odyssey4me: Right. No rush, like I said we know the root cause and we can always push a doc patch to begin with. | 11:40 |
| fghaas | Then consider options for how to fix sudo invocation, if we want to. | 11:41 |
| odyssey4me | hmm, ok - I think I see the issue here | 11:45 |
| odyssey4me | ansible_env.HOME is the running user's home, but ansible_user_id is root, as is ansible_env.USER | 11:47 |
| odyssey4me | ansible_env.HOME is the running user's home, but ansible_user_id is root, as is ansible_env.USER - we could try using ansible_env.SUDO_USER if it is available, and fall back to ansible_env.USER | 11:48 |
| openstackgerrit | Merged openstack/openstack-ansible-os_glance stable/rocky: Make glance cache management cron task idempotent https://review.openstack.org/612752 | 11:49 |
| openstackgerrit | Merged openstack/openstack-ansible-plugins stable/rocky: Allow basic auth in netloc_no_port filtered URLs https://review.openstack.org/614840 | 11:52 |
| *** cshen has quit IRC | 11:59 | |
| *** cshen has joined #openstack-ansible | 12:04 | |
| odyssey4me | hwoarang I'm booting up a vagrant box now to see what's going on with http://logs.openstack.org/77/614777/1/check/openstack-ansible-deploy-aio_metal-opensuse-423/1a94cb3/logs/ara-report/result/2272ec6b-aa44-4824-b6d4-4c82f9f0488a/ | 12:05 |
| hwoarang | ok odyssey4me. let me know if i can be of any help | 12:06 |
| * hwoarang is looking at some ceph+leap15 issues right now | 12:06 | |
| *** chhavi__ has quit IRC | 12:38 | |
| *** maddtux has quit IRC | 12:41 | |
| vollman | odyssey4me: Do have time to take a look at this one? https://review.openstack.org/#/c/614210 | 12:56 |
| *** bertys has joined #openstack-ansible | 12:56 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_nova master: Retry waiting for initial nova-compute with rescue https://review.openstack.org/614650 | 12:56 |
| bertys | hi cores - may I draw again your attention to https://review.openstack.org/#/c/549956/ ? Thanks in advance for your feedback | 13:00 |
| *** strattao has joined #openstack-ansible | 13:07 | |
| *** faizy98 has joined #openstack-ansible | 13:10 | |
| fghaas | odyssey4me: yes exactly. ansible_user_id gives you the euid, however ansible_env.HOME is, well, $HOME, and since sudo doesn't reset that unless invoked with -i or -H, that's where the trouble starts. | 13:10 |
| *** faizy_ has quit IRC | 13:12 | |
| *** vakuznet has joined #openstack-ansible | 13:18 | |
| vakuznet | odyssey4me could you take a look https://review.openstack.org/#/c/612011/ | 13:19 |
| hwoarang | bertys: looks good to me | 13:24 |
| hwoarang | vakuznet: approved based on the previous discussion since i am not familiar with the octavia role | 13:27 |
| vakuznet | thanks | 13:29 |
| openstackgerrit | Jimmy McCrory proposed openstack/ansible-role-systemd_service master: Allow services to provide after_targets directly https://review.openstack.org/615185 | 13:33 |
| jrosser | odyssey4me: fghaas if you add a “must be root” bailout to openstack-ansible wrapper please make it possible to opt out of the check somehow | 13:46 |
| *** aredinger has joined #openstack-ansible | 13:46 | |
| fghaas | jrosser: no, for the time being I lean just towards a doc patch | 13:46 |
| openstackgerrit | Jimmy McCrory proposed openstack/openstack-ansible-os_nova master: Avoid looping systemd_service role include https://review.openstack.org/615192 | 13:51 |
| *** spatel has joined #openstack-ansible | 14:01 | |
| fghaas | jrosser: out of curiosity, have you done a rocky deployment with ceph-osd hosts where osd_objectstore was set to bluestore? | 14:06 |
| openstackgerrit | Michael Vollman proposed openstack/openstack-ansible-ops master: Fix ansible path for elk_metrics_6x tests https://review.openstack.org/614210 | 14:07 |
| dcdamien | odyssey4me: open-iscsi related proposal ready and pending review: https://review.openstack.org/#/c/614556/ | 14:11 |
| *** dcdamien has quit IRC | 14:15 | |
| *** dcdamien has joined #openstack-ansible | 14:15 | |
| *** evrardjp_ has joined #openstack-ansible | 14:20 | |
| jrosser | fghaas: i upgraded my lab Q -> R yesterday, however my deployments are always with OSA consuming an already existing ceph cluster | 14:28 |
| openstackgerrit | Merged openstack/openstack-ansible-os_tempest master: Ubuntu distro install uses testrepository https://review.openstack.org/614888 | 14:28 |
| jrosser | but that ceph cluster is mimc/bluestore currently | 14:28 |
| fghaas | right, I was (a) wondering why we still default to filestore, and (b) I just deployed a Luminous test cluster with osd_objectstore set to bluestore and my OSDs wouldn't peer. | 14:29 |
| jrosser | in truth, i can't keep all the moving parts in my head with the OSA/ceph combined deployment | 14:30 |
| fghaas | :) | 14:30 |
| fghaas | do you run ceph-osd and nova-compute on the same nodes? | 14:30 |
| jrosser | so it's easier for me/my team to treat it as two seperate things, and theres a very small bit of data (the mon/rgw ips) and some auth keys to share between the two | 14:31 |
| jrosser | my ceph and openstack hosts are entirely seperate | 14:31 |
| jrosser | avoids two possibly contradictory sets of deployment tooling trampling on each other | 14:32 |
| logan- | fghaas: do we set filestore in OSA? I think that might be the upstream default still :/ | 14:32 |
| odyssey4me | fghaas I know that logan- does a combined deployment, and is using bluestore. | 14:32 |
| fghaas | logan-: you mean "upstream" as in ceph-ansible, not Ceph proper? | 14:33 |
| logan- | correct ceph-ansible | 14:33 |
| fghaas | huh. Let me go check. | 14:33 |
| logan- | i don't think we override whatever their default is, but i thought I saw a PR a few months ago changing it to bluestore | 14:33 |
| fghaas | Their defaults sometimes confusingly change with releases | 14:33 |
| fghaas | (and we don't track stable branches in OSA, as mentioned here yesterday) | 14:34 |
| fghaas | logan-, jrosser: confirmed, filestore in stable-3.1, bluestore in stable-3.2 | 14:35 |
| fghaas | the problem with that is, I think, that that means in OSA stable/rocky we just flipped that default without so much as a reno | 14:36 |
| fghaas | (by just refreshing the sha for ceph-ansible) | 14:36 |
| logan- | yeah we should definitely reno the ceph-ansible major version bumps, maybe not every detail of it but a pointer to some appropriate upstream changelog or something | 14:37 |
| jrosser | i mentioned similar yesterday about the var setting the ceph release | 14:37 |
| logan- | bear in mind the filestore->bluestore change will only affect new osd preparations, it won't touch the existing filestore osds | 14:37 |
| jrosser | i'm not sure that is necessarily meaningful with all the combinations of distro/uca/ceph.com packages vs. centos/suse/ubuntu release that folks can choose from now | 14:38 |
| fghaas | I am acutely aware of that, but still you'll get massive performance discrepancies if you just sprinkle bluestore OSDs into a filestore cluster willy-nilly | 14:38 |
| logan- | yeah | 14:38 |
| openstackgerrit | Merged openstack/openstack-ansible-os_octavia master: octavia container network address pool https://review.openstack.org/612011 | 14:47 |
| openstackgerrit | Merged openstack/openstack-ansible-os_glance stable/rocky: zuul: Switch the Xenial distro job to Bionic https://review.openstack.org/615116 | 14:54 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-tests master: [DNM] - Testing debug log collection https://review.openstack.org/581642 | 15:01 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build master: Ensure pip/setuptools/wheel are up to date in the wheel build venv https://review.openstack.org/615205 | 15:02 |
| odyssey4me | hwoarang ^ that fixes one issue I encountered, which will need porting back to rocky | 15:03 |
| hwoarang | interesting patch... let me understand it | 15:05 |
| odyssey4me | the wheel build failed at cryptography which needs a newer version of setuptools which is placed there by virtualenv | 15:06 |
| hwoarang | ok i see | 15:07 |
| odyssey4me | this is pretty much the same as was done in https://github.com/openstack/openstack-ansible-os_gnocchi/commit/6a9370ed5d1cdff580c74c24a600eb4393d119c4 | 15:08 |
| odyssey4me | we may unfortunately end up having to do this in a lot of roles | 15:08 |
| *** bertys has quit IRC | 15:11 | |
| *** cshen has quit IRC | 15:12 | |
| *** ThiagoCMC has joined #openstack-ansible | 15:12 | |
| *** chandankumar is now known as chkumar|off | 15:14 | |
| *** rgogunskiy has quit IRC | 15:15 | |
| *** cshen has joined #openstack-ansible | 15:16 | |
| spotz | Anyone still needing a hangout invite for Berlin please DM me to get you added, we're already doing plotting:) | 15:18 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build stable/rocky: Use --never-download instead of --no-download https://review.openstack.org/615213 | 15:20 |
| odyssey4me | hwoarang ^ another one | 15:20 |
| odyssey4me | We don't see that in gates because infra puts a new version in the base image. | 15:20 |
| spotz | odyssey4me: So just changed from --no-download to --never-download? | 15:24 |
| odyssey4me | spotz yep, --no-download requires a newer version of virtualenv which is not present in suse 42 | 15:24 |
| spotz | odyssey4me: Ha, well if we need that change pushed out multiple places let me know! My favorite kind of patch:) | 15:25 |
| *** dave-mccowan has quit IRC | 15:25 | |
| odyssey4me | spotz it shouldn't be needed elsewhere | 15:25 |
| odyssey4me | but thanks for the offer :) | 15:25 |
| *** dave-mccowan has joined #openstack-ansible | 15:26 | |
| *** gyee has joined #openstack-ansible | 15:26 | |
| *** cshen has quit IRC | 15:29 | |
| spotz | odyssey4me I'm strange I like the tedious and mundane:) | 15:30 |
| *** sum12 has quit IRC | 15:46 | |
| openstackgerrit | Antony Messerli proposed openstack/openstack-ansible master: Adds option to enable ara during upgrades https://review.openstack.org/612775 | 15:46 |
| *** vakuznet has quit IRC | 15:46 | |
| openstackgerrit | Antony Messerli proposed openstack/openstack-ansible master: Adds ara during upgrades https://review.openstack.org/612775 | 15:48 |
| *** sum12 has joined #openstack-ansible | 15:48 | |
| openstackgerrit | Jimmy McCrory proposed openstack/openstack-ansible master: Update ansible to 2.7.1 https://review.openstack.org/615225 | 15:48 |
| *** blinkiz has joined #openstack-ansible | 16:01 | |
| *** blinkiz has left #openstack-ansible | 16:01 | |
| *** strattao has quit IRC | 16:04 | |
| *** strattao has joined #openstack-ansible | 16:09 | |
| *** sum12 has quit IRC | 16:16 | |
| *** sum12 has joined #openstack-ansible | 16:16 | |
| *** pcaruana has quit IRC | 16:17 | |
| *** cshen has joined #openstack-ansible | 16:26 | |
| *** fghaas has quit IRC | 16:31 | |
| openstackgerrit | Merged openstack/openstack-ansible stable/ocata: Update Ocata SHA for Keystone memcache flush https://review.openstack.org/613956 | 16:32 |
| *** ThiagoCMC has quit IRC | 16:32 | |
| *** strattao has quit IRC | 16:37 | |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: [DNM] Demonstrate the use of Mitogen with OSA https://review.openstack.org/591236 | 16:38 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: [DNM] Demonstrate the use of Mitogen with OSA https://review.openstack.org/591236 | 16:38 |
| dcdamien | odyssey4me: You've introduced this https://review.openstack.org/615213. I'm doing Q -> R upgrade. On keystone playbook when it's checking if service is already available (using URI module) it fails because it delegates to deployment host which doesn't have resolver set to unbound (i'm using dns name as internal URL). Isn't it some kind of logical bug? | 16:59 |
| *** DanyC has quit IRC | 17:00 | |
| *** DanyC has joined #openstack-ansible | 17:03 | |
| *** DanyC has quit IRC | 17:04 | |
| dcdamien | odyseey4me: Why we are not defaulting openstack_service_setup_host to utility container? | 17:05 |
| odyssey4me | dcdamien because you may not have one - it's not required | 17:08 |
| odyssey4me | you definitely need a deployment host | 17:08 |
| odyssey4me | dcdamien which is the task that's failing? the review you linked is totally unrelated | 17:09 |
| dcdamien | Ah, sorry. One sec | 17:10 |
| logan- | seems like youd run into the same issue if you were using a dns name for your internal vip without unbound too, since i doubt we run the hosts file syncing on localhost | 17:11 |
| odyssey4me | are you talking about this one? https://github.com/openstack/openstack-ansible-os_keystone/blob/master/tasks/keystone_service_update.yml#L32 | 17:11 |
| logan- | though i guess the internal vip might more likely to be publicly resolvable in that case | 17:12 |
| dcdamien | yeah, this fails with default service setup host in my case when i'm using dns name for internal VIP | 17:12 |
| odyssey4me | dcdamien in that case it will expect that the service setup host can resolve the name | 17:12 |
| dcdamien | ok, maybe we just need to highlight it in documentation | 17:13 |
| openstackgerrit | Merged openstack/openstack-ansible-repo_build stable/rocky: Use --never-download instead of --no-download https://review.openstack.org/615213 | 17:13 |
| jrosser | odyssey4me: if the utility container is optional you might have a better idea of what to do here https://review.openstack.org/#/c/614591/ | 17:14 |
| jrosser | perhaps openstack_service_setup_host instead? | 17:14 |
| dcdamien | jrosser: your issue looks pretty similar :D | 17:15 |
| jrosser | same end result, different root cause i think | 17:16 |
| odyssey4me | jrosser that play, and all the other bits around controlling the galera upgrade in the upgrade script pre-dates the implementation of serialisation in https://github.com/openstack/openstack-ansible/blob/master/playbooks/galera-install.yml#L19 | 17:17 |
| dcdamien | In my case it waits for keystone to be available as I see you wait for MariaDB. Both requires connectivity (and resolvability). | 17:17 |
| odyssey4me | jrosser it's highly likely that much of that can just be removed nowadays | 17:18 |
| dcdamien | odyssey4me: isn't better option to choose utility if available with fallback to localhost as default? | 17:20 |
| odyssey4me | dcdamien I don't think so. If that's what you'd like, then it's a single variable to set. | 17:21 |
| *** hamzaachi has joined #openstack-ansible | 17:24 | |
| mnaser | odyssey4me: so with python_venv_builder, repo_server is mostly just a pypi cache at this point? | 17:25 |
| odyssey4me | mnaser yes-ish, slowly moving towards making that the case in stein | 17:26 |
| mnaser | odyssey4me: ok gotcha, so really our installs should speed up way more because most dependencies are most likely pulled in from pypi/cache | 17:26 |
| odyssey4me | mnaser they always have been - this work is not about improving speed, it's about making it easier to understand | 17:27 |
| odyssey4me | it may speed them up marginally because the build will be spread over the roles, not all done at once | 17:27 |
| mnaser | odyssey4me: ah okay, i thought we built wheels before but i assume that was just openstack wheels only | 17:27 |
| odyssey4me | so therell be less io contention | 17:28 |
| odyssey4me | nope, we built all required python packages into wheels in the repo_build, then packaged them into venvs, then each role pulled down the venv and unpacked it | 17:28 |
| odyssey4me | right now we've pulled out the venv packaging because it was getting harder and harder to maintain | 17:29 |
| odyssey4me | we're still always building wheels and consuming them | 17:29 |
| odyssey4me | but once repo build is gone, the wheels will get built on demand | 17:29 |
| mnaser | ah so that was the "stage 2" you were talking about | 17:29 |
| odyssey4me | yep - next step right now is to eliminate the pip.conf so that all instructions are clearly in the task, not hidden in a file | 17:30 |
| odyssey4me | to do that we need to pull all the constraints into the roles - and I'm working on that | 17:30 |
| mnaser | odyssey4me: cool, gotcha. on a seperate note, do you remember a role on the top of your head where we used jmespath instead of having 6 lines of when: | 17:31 |
| odyssey4me | given that we build from git SHA's, it makes it a little more complex than I'd like - but that's the brief | 17:31 |
| odyssey4me | I don't. :/ | 17:31 |
| mnaser | ok ill try to remember where to find an example | 17:32 |
| jrosser | mnaser: not sure if this is the one you mean but..... https://github.com/openstack/openstack-ansible/commit/6d9f7f23372cc607a5f624c0d9407d5b752e8ae1#diff-255f0b55f1833e385fc3b3edbe740135 | 17:35 |
| spotz | mnaser openstack-ansible-ops openstack-ansible openstack-ansible-repo_build openstack-ansible-tests all have hits on jmespath when using codesearch | 17:37 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_neutron master: Drop legacy HA tool from being installed https://review.openstack.org/615255 | 17:40 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-plugins master: Drop name2int as it's no longer used https://review.openstack.org/615256 | 17:41 |
| mnaser | spotz: ah good idea | 17:42 |
| spotz | mnaser: We defitinely have a version requirement in those | 17:42 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_neutron master: Drop legacy HA tool from being installed https://review.openstack.org/615255 | 17:43 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/rocky: Ensure pip/setuptools/wheel are up to date in the utility venv https://review.openstack.org/615258 | 17:44 |
| odyssey4me | hwoarang ^ yet another one crawled out of the woodwork | 17:44 |
| odyssey4me | mnaser could you review that one and https://review.openstack.org/#/c/615205/1 too please | 17:45 |
| mnaser | oh | 17:46 |
| mnaser | i like the rescue thing | 17:46 |
| odyssey4me | I'm working through a vagrant build of rocky on suse 42 and hitting all sorts of broken, which is good because we get to fix them. | 17:46 |
| mnaser | instead of a bunch of tasks | 17:46 |
| guilhermesp | hello all! Seems that we have os_placement repo in place https://review.openstack.org/#/admin/projects/openstack/openstack-ansible-os_placement | 17:49 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_neutron master: Refactor pre-install checks into neutron_check.yml https://review.openstack.org/615262 | 17:49 |
| guilhermesp | now it's time to begin the development | 17:49 |
| odyssey4me | awesome, thanks for picking that up guilhermesp ! | 17:50 |
| guilhermesp | np odyssey4me ! just wondering, I have a really basic directory/files structure for the role. Would be good to create an initial PR for this? | 17:50 |
| mnaser | guilhermesp: i say you can just have one big commit that includes a .zuul.yaml file -- that way we can test it as we merge it the first time | 17:51 |
| odyssey4me | guilhermesp Sure, reviews are much easier to look through when they're in smaller parts | 17:51 |
| odyssey4me | if you get the scaffolding in first, and add the basic tests for linters, docs, releasenotes, etc - but leave out functional tests for later | 17:52 |
| guilhermesp | cool thanks odyssey4me and mnaser this is the skeleton that I built till now http://paste.openstack.org/show/734050/ | 17:53 |
| odyssey4me | I'm out for the night folks, and probably the w/end. Have a great one! | 17:53 |
| mnaser | later odyssey4me ! | 17:53 |
| guilhermesp | I was using this as I reference http://blog.andy.mc.it/osa-role-pt1/ | 17:53 |
| guilhermesp | have a nice weekend odyssey4me ! | 17:53 |
| odyssey4me | guilhermesp that's a bit dated - make sure you look at changes in the current head of one of the simpler roles that's well maintained - like cinder/glance | 17:54 |
| guilhermesp | yep, I was using os_glance as reference | 17:54 |
| guilhermesp | I'm still using btw | 17:54 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/rocky: Ensure pip/setuptools/wheel are up to date in the utility venv https://review.openstack.org/615258 | 17:54 |
| guilhermesp | mnaser: when you say .zuul.yaml it is something like this https://github.com/openstack/placement/blob/master/.zuul.yaml , isn't it ? | 18:06 |
| guilhermesp | without functional tests for now | 18:06 |
| mnaser | guilhermesp: yes but you should use the zuul file from roles like os_glance | 18:06 |
| mnaser | not use the project one | 18:07 |
| guilhermesp | yes, I used that link because I couldn't find .zuul.yaml in os_glance repo | 18:07 |
| mnaser | i think its in zuul.d/project.yaml | 18:07 |
| -openstackstatus- NOTICE: OpenStack infra's mirror nodes stopped accepting connections on ports 8080, 8081, and 8082. We will notify when this is fixed and jobs can be rechecked if they failed to communicate with a mirror on these ports. | 18:08 | |
| guilhermesp | yes, that one I was wondering | 18:08 |
| openstackgerrit | Guilherme Steinmuller Pimentel proposed openstack/openstack-ansible-os_placement master: Initial commit to os_placement role https://review.openstack.org/615272 | 18:15 |
| guilhermesp | mnaser: just a kick off ^ | 18:15 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_nova master: Avoid including nova_virt_detect if nova_virt_type is set https://review.openstack.org/615274 | 18:17 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_nova master: Drop old placement NGINX code https://review.openstack.org/615275 | 18:17 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_nova master: Refactor libvirt default network remove task https://review.openstack.org/615276 | 18:17 |
| mnaser | and that explains our failures | 18:22 |
| mnaser | guilhermesp there's a few references still of glance | 18:22 |
| guilhermesp | going to change | 18:22 |
| openstackgerrit | Guilherme Steinmuller Pimentel proposed openstack/openstack-ansible-os_placement master: Initial commit to os_placement role https://review.openstack.org/615272 | 18:24 |
| *** electrofelix has quit IRC | 18:24 | |
| *** pcaruana has joined #openstack-ansible | 18:27 | |
| *** chhavi__ has joined #openstack-ansible | 18:34 | |
| *** chhavi__ has quit IRC | 18:39 | |
| openstackgerrit | Guilherme Steinmuller Pimentel proposed openstack/openstack-ansible-os_placement master: Initial commit to os_placement role https://review.openstack.org/615272 | 18:50 |
| *** mmercer has joined #openstack-ansible | 18:51 | |
| -openstackstatus- NOTICE: The firewall situation with ports 8080, 8081, and 8082 on mirror nodes has been resolved. You can recheck jobs that have failed to communicate to the mirrors on those ports now. | 18:52 | |
| *** pcaruana has quit IRC | 19:08 | |
| openstackgerrit | Guilherme Steinmuller Pimentel proposed openstack/openstack-ansible-os_placement master: Initial commit to os_placement role https://review.openstack.org/615272 | 19:19 |
| *** fghaas has joined #openstack-ansible | 19:29 | |
| *** strattao has joined #openstack-ansible | 19:31 | |
| jrosser | ooooh theres the first build of any kind to complete successfully with mitogen https://review.openstack.org/#/c/591236/ | 20:02 |
| jrosser | the prise goes to opensuse | 20:02 |
| jrosser | *prize | 20:02 |
| spotz | Woot! | 20:18 |
| *** hamzaachi has quit IRC | 20:23 | |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_keystone master: Drop deprecated memcached settings https://review.openstack.org/615308 | 20:46 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_keystone master: Drop UUID token tasks https://review.openstack.org/615309 | 20:46 |
| *** spatel has quit IRC | 20:49 | |
| *** fghaas has quit IRC | 20:57 | |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-os_swift master: Correct swift replication service units https://review.openstack.org/615310 | 20:57 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_nova master: Drop un-used nova_nginx_conf_path https://review.openstack.org/615312 | 20:59 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_nova master: Remove un-used nova_compute_manager https://review.openstack.org/615313 | 21:08 |
| jrosser | mnaser: http://logs.openstack.org/36/591236/5/check/openstack-ansible-deploy-aio_metal-ubuntu-bionic/3edfba4/logs/ara-report/ | 21:13 |
| jrosser | 40mins! | 21:14 |
| mnaser | what type of black magic | 21:14 |
| jrosser | perhaps theres a node with some magical afterburner on it and that job got lucky | 21:14 |
| jrosser | it's a metal job, so no containers which is some of it | 21:15 |
| mnaser | still... | 21:15 |
| jrosser | yeah, i randomly looked up another of the same kind which was 1hr10 | 21:16 |
| mnaser | openstack-ansible-deploy-aio_metal-ubuntu-bionic | 21:16 |
| mnaser | 1h04m | 21:16 |
| mnaser | yeah | 21:16 |
| mnaser | is there a reason its using your own branch right now? | 21:16 |
| jrosser | i have a small but necessary patch to mitogen | 21:17 |
| mnaser | i see | 21:18 |
| mnaser | is it upstreamable? | 21:18 |
| jrosser | we should look at whats going on there - someone who understands the connection plugin / inventory better than me would be useful | 21:18 |
| jrosser | it used the wrong data, ip rather than container name when using lxc-attach / nsenter | 21:19 |
| jrosser | so i just dumped the dict in mitogen and changed it that to use a var that had the right value | 21:19 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_neutron master: Switch to openstack-ansible-role-distro_install-jobs https://review.openstack.org/615318 | 21:20 |
| *** strattao has quit IRC | 21:20 | |
| jrosser | otoh, the metal deploy won't actually use any of that..... | 21:20 |
| mnaser | ah we probably need to make some dynamic_inventory changes then | 21:20 |
| guilhermesp | is there any thing I'm missing here? http://logs.openstack.org/72/615272/4/check/openstack-tox-docs/44e24f7/job-output.txt.gz#_2018-11-02_19_21_50_384007 | 21:20 |
| mnaser | that is a weird error guilhermesp | 21:22 |
| mnaser | but i think that error is ok? | 21:22 |
| mnaser | http://logs.openstack.org/72/615272/4/check/openstack-tox-docs/44e24f7/job-output.txt.gz#_2018-11-02_19_21_54_438610 | 21:22 |
| mnaser | this is what is failing | 21:22 |
| guilhermesp | hum yeah, probably the title | 21:24 |
| guilhermesp | let me fix it | 21:25 |
| openstackgerrit | Guilherme Steinmuller Pimentel proposed openstack/openstack-ansible-os_placement master: Initial commit to os_placement role https://review.openstack.org/615272 | 21:25 |
| mnaser | guilhermesp: added a few comments too | 21:27 |
| jrosser | mnaser: http://logs.openstack.org/36/591236/5/check/openstack-ansible-deploy-aio_metal-ubuntu-bionic/3edfba4/zuul-info/host-info.ubuntu-bionic.yaml | 21:27 |
| mnaser | jrosser: of course ;) | 21:28 |
| jrosser | :) | 21:28 |
| *** fghaas has joined #openstack-ansible | 21:29 | |
| *** dave-mccowan has quit IRC | 21:35 | |
| openstackgerrit | Guilherme Steinmuller Pimentel proposed openstack/openstack-ansible-os_placement master: Initial commit to os_placement role https://review.openstack.org/615272 | 21:38 |
| guilhermesp | done mnaser ^ | 21:38 |
| *** fghaas has quit IRC | 21:39 | |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-openstack_hosts master: refactor: remove the blacklisted packages https://review.openstack.org/615335 | 21:39 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-openstack_hosts master: refactor: remove unused proxy settings file https://review.openstack.org/615337 | 21:41 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-openstack_hosts master: refactor: never set no-cache https://review.openstack.org/615338 | 21:43 |
| *** DanyC has joined #openstack-ansible | 21:55 | |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-lxc_hosts master: Drop COPR priority settings https://review.openstack.org/615349 | 21:58 |
| guilhermesp | all green for now https://review.openstack.org/#/c/615272/ thanks mnaser | 21:59 |
| *** DanyC has quit IRC | 21:59 | |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-lxc_hosts master: lxc: stop mirroring COPR locally on CentOS 7 https://review.openstack.org/615351 | 22:05 |
| *** aredinger has quit IRC | 22:35 | |
| *** spatel has joined #openstack-ansible | 22:46 | |
| *** spatel has quit IRC | 22:50 | |
| *** aredinger has joined #openstack-ansible | 22:54 | |
| *** aredinger has quit IRC | 22:59 | |
| *** ianychoi has quit IRC | 23:02 | |
| *** shardy_ has quit IRC | 23:18 | |
| *** shardy has quit IRC | 23:18 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_magnum stable/queens: Add multi-region support for Magnum https://review.openstack.org/610324 | 23:32 |
| *** markvoelker has quit IRC | 23:41 | |
| *** markvoelker has joined #openstack-ansible | 23:41 | |
| *** markvoelker has quit IRC | 23:46 | |
| *** gyee has quit IRC | 23:48 | |
| *** aredinger has joined #openstack-ansible | 23:54 | |
| *** ianychoi has joined #openstack-ansible | 23:55 | |
| *** aredinger has quit IRC | 23:59 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!