| *** markvoelker has joined #openstack-ansible | 00:09 | |
| *** nurdie has joined #openstack-ansible | 00:33 | |
| *** nurdie_ has joined #openstack-ansible | 00:34 | |
| *** nurdie has quit IRC | 00:37 | |
| *** markvoelker has quit IRC | 00:42 | |
| *** markvoelker has joined #openstack-ansible | 00:43 | |
| *** nurdie has joined #openstack-ansible | 00:44 | |
| *** nurdie_ has quit IRC | 00:48 | |
| *** nurdie has quit IRC | 00:49 | |
| *** gyee has quit IRC | 00:49 | |
| *** markvoelker has quit IRC | 01:37 | |
| *** markvoelker has joined #openstack-ansible | 01:44 | |
| *** jhesketh has quit IRC | 01:53 | |
| *** joshualyle has joined #openstack-ansible | 02:14 | |
| *** joshualyle has quit IRC | 02:18 | |
| *** jhesketh has joined #openstack-ansible | 02:22 | |
| *** rholloway has quit IRC | 03:10 | |
| *** BjoernT has joined #openstack-ansible | 03:17 | |
| *** BjoernT has quit IRC | 03:21 | |
| *** BjoernT has joined #openstack-ansible | 03:21 | |
| *** jhesketh has quit IRC | 03:29 | |
| *** gkadam has joined #openstack-ansible | 03:31 | |
| *** jhesketh has joined #openstack-ansible | 03:36 | |
| *** gkadam has quit IRC | 03:40 | |
| *** raukadah is now known as chkumar|ruck | 03:49 | |
| *** udesale has joined #openstack-ansible | 04:09 | |
| *** BjoernT has quit IRC | 04:17 | |
| *** BjoernT has joined #openstack-ansible | 04:26 | |
| *** dr_feelgood has joined #openstack-ansible | 04:56 | |
| *** shyamb has joined #openstack-ansible | 05:01 | |
| *** spsurya has joined #openstack-ansible | 05:02 | |
| *** shyamb has quit IRC | 05:04 | |
| *** shyamb has joined #openstack-ansible | 05:05 | |
| *** shyam89 has joined #openstack-ansible | 05:07 | |
| *** shyamb has quit IRC | 05:11 | |
| *** nurdie has joined #openstack-ansible | 05:13 | |
| *** BjoernT has quit IRC | 05:15 | |
| *** nurdie has quit IRC | 05:18 | |
| *** shyamb has joined #openstack-ansible | 05:43 | |
| *** shyam89 has quit IRC | 05:43 | |
| *** shyamb has quit IRC | 05:54 | |
| *** shyam89 has joined #openstack-ansible | 05:54 | |
| *** nurdie has joined #openstack-ansible | 05:54 | |
| *** nurdie has quit IRC | 06:00 | |
| *** nurdie has joined #openstack-ansible | 06:15 | |
| *** nurdie has quit IRC | 06:20 | |
| *** dr_feelgood has quit IRC | 06:21 | |
| *** jbadiapa has joined #openstack-ansible | 06:33 | |
| *** nurdie has joined #openstack-ansible | 06:36 | |
| *** udesale has quit IRC | 06:41 | |
| *** nurdie has quit IRC | 06:41 | |
| *** udesale has joined #openstack-ansible | 06:42 | |
| *** nurdie has joined #openstack-ansible | 06:57 | |
| *** miloa has joined #openstack-ansible | 06:57 | |
| *** ianychoi has quit IRC | 07:01 | |
| *** ianychoi has joined #openstack-ansible | 07:01 | |
| *** nurdie has quit IRC | 07:02 | |
| *** udesale has quit IRC | 07:03 | |
| *** udesale has joined #openstack-ansible | 07:03 | |
| *** devx has quit IRC | 07:07 | |
| *** devx has joined #openstack-ansible | 07:09 | |
| miloa | Morning | 07:14 |
|---|---|---|
| jrosser | Morning | 07:14 |
| *** tosky has joined #openstack-ansible | 07:17 | |
| *** nurdie has joined #openstack-ansible | 07:18 | |
| *** nurdie has quit IRC | 07:23 | |
| *** ianychoi has quit IRC | 07:24 | |
| *** shyam89 has quit IRC | 07:25 | |
| *** ianychoi has joined #openstack-ansible | 07:25 | |
| *** shyam89 has joined #openstack-ansible | 07:25 | |
| *** udesale has quit IRC | 07:57 | |
| *** udesale has joined #openstack-ansible | 07:58 | |
| *** nurdie has joined #openstack-ansible | 07:59 | |
| *** kopecmartin|off is now known as kopecmartin | 07:59 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_ceilometer stable/stein: Change default username for oslo from designate to designate-rpc https://review.opendev.org/675088 | 08:01 |
| *** nurdie has quit IRC | 08:04 | |
| *** rpittau|afk is now known as rpittau | 08:04 | |
| *** ivve has joined #openstack-ansible | 08:19 | |
| *** shyamb has joined #openstack-ansible | 08:25 | |
| *** shyam89 has quit IRC | 08:27 | |
| *** shyamb has quit IRC | 08:33 | |
| *** jbadiapa has quit IRC | 08:36 | |
| *** nurdie has joined #openstack-ansible | 08:40 | |
| *** nurdie has quit IRC | 08:45 | |
| *** Adri2000 has quit IRC | 08:53 | |
| openstackgerrit | Erik Berg proposed openstack/openstack-ansible-os_neutron master: network_hosts do dvr_snat, everything else can (probably) do dvr https://review.opendev.org/675274 | 08:57 |
| *** nurdie has joined #openstack-ansible | 09:01 | |
| *** nurdie has quit IRC | 09:06 | |
| *** electrofelix has joined #openstack-ansible | 09:12 | |
| *** nurdie has joined #openstack-ansible | 09:42 | |
| *** shyamb has joined #openstack-ansible | 09:43 | |
| *** nurdie has quit IRC | 09:48 | |
| *** udesale has quit IRC | 10:01 | |
| *** udesale has joined #openstack-ansible | 10:01 | |
| *** udesale has quit IRC | 10:02 | |
| *** udesale has joined #openstack-ansible | 10:03 | |
| *** alexusono has left #openstack-ansible | 10:06 | |
| *** Adri2000 has joined #openstack-ansible | 10:08 | |
| *** shyamb has quit IRC | 10:19 | |
| *** shyamb has joined #openstack-ansible | 10:19 | |
| *** kdean has quit IRC | 10:21 | |
| noonedeadpunk | morning folks | 10:23 |
| *** nurdie has joined #openstack-ansible | 10:24 | |
| noonedeadpunk | wow, thanks jrosser for reviews | 10:26 |
| *** nurdie has quit IRC | 10:29 | |
| noonedeadpunk | So I guess https://review.opendev.org/#/c/674649/7 is pretty safe in terms why we are reverting smart sources - since it do not symlink /etc/service folder (it was the problem iirc) | 10:30 |
| noonedeadpunk | It just retrieves files from venv, pass it trough config_tempate to apply overrides and drop to /etc/service which in general should be ok... It's only for source installs though - I guess required files should present in packages | 10:32 |
| *** udesale has quit IRC | 10:38 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_tacker master: service_setup: refactor service setup to a single file https://review.opendev.org/675289 | 10:40 |
| *** nurdie has joined #openstack-ansible | 10:45 | |
| *** nurdie has quit IRC | 10:49 | |
| *** shyamb has quit IRC | 11:00 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_trove master: service_setup: refactor service setup to a single file https://review.opendev.org/675293 | 11:10 |
| *** kplant has joined #openstack-ansible | 11:18 | |
| *** shyamb has joined #openstack-ansible | 11:22 | |
| *** ansmith has quit IRC | 11:25 | |
| *** nurdie has joined #openstack-ansible | 11:26 | |
| *** nurdie has quit IRC | 11:31 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_zun master: service_setup: refactor service setup to a single file https://review.opendev.org/675300 | 11:32 |
| *** shyamb has quit IRC | 12:13 | |
| *** shyamb has joined #openstack-ansible | 12:14 | |
| *** shyamb has quit IRC | 12:16 | |
| *** shyam89 has joined #openstack-ansible | 12:16 | |
| *** shyam89 has quit IRC | 12:21 | |
| *** udesale has joined #openstack-ansible | 12:28 | |
| *** ansmith has joined #openstack-ansible | 12:37 | |
| mnaser | hi all | 12:50 |
| mgariepy | morning mnaser | 12:51 |
| openstackgerrit | Merged openstack/openstack-ansible-os_manila master: Fix db_setup condition https://review.opendev.org/673343 | 12:52 |
| openstackgerrit | Merged openstack/openstack-ansible-os_manila master: Set defaults for project-specific variables https://review.opendev.org/673350 | 12:57 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_manila master: smart_sources: Use config files from repo https://review.opendev.org/674649 | 13:00 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_manila master: Add variable files for RedHat and Suse https://review.opendev.org/674851 | 13:00 |
| *** rholloway has joined #openstack-ansible | 13:24 | |
| *** jcosmao has joined #openstack-ansible | 13:27 | |
| *** BjoernT has joined #openstack-ansible | 13:28 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_cloudkitty master: Replace include with import https://review.opendev.org/675092 | 13:34 |
| noonedeadpunk | jrosser: yestarday you said that we somehow adding source packages for ubuntu distro installs... I just can't find anything reated so for me it looks like we just don't install any plugins for ubuntu in case of distro installs... | 13:38 |
| noonedeadpunk | it's about tempest:) | 13:38 |
| jrosser | noonedeadpunk: i read this https://github.com/openstack/openstack-ansible-os_tempest/blob/master/vars/debian.yml#L32-L36 | 13:45 |
| noonedeadpunk | Yeah, I aready read... | 13:45 |
| jrosser | and then i wondered how thqt worked and a quick look didnt really help me | 13:45 |
| *** KeithMnemonic has quit IRC | 13:45 | |
| noonedeadpunk | Ah, ok:) | 13:45 |
| jrosser | like where do the tempest plugins come from at all with ubuntu distro, i didnt see it quickly | 13:45 |
| *** KeithMnemonic has joined #openstack-ansible | 13:45 | |
| noonedeadpunk | so I guess tempest distro package just contains some plugins... | 13:45 |
| jrosser | it could well contain a base set yes | 13:46 |
| jrosser | i think there was a previous attempt to make the tempest role sort of half source half distro | 13:46 |
| jrosser | but i'm not sure if that ever merged | 13:46 |
| noonedeadpunk | So it seems that it's not... | 13:48 |
| noonedeadpunk | And python-tempest contains some basic tests, so that's why we're probably ok https://packages.ubuntu.com/bionic/all/python-tempest/filelist | 13:48 |
| noonedeadpunk | So for mixed install I think we need to install plugins without venv to standart python | 13:50 |
| noonedeadpunk | And I feel that I won't like result in terms of methodics and code beauty... | 13:52 |
| *** electrofelix has quit IRC | 13:57 | |
| jrosser | the issue was trying to use install method as both source and distro setting in different places in the code | 13:57 |
| jrosser | perhaps it needs another var for distro-with-source-plugins | 13:58 |
| *** dave-mccowan has joined #openstack-ansible | 13:58 | |
| noonedeadpunk | in terms of copyright and this stuff - can I just copy odyssey4me's comment https://github.com/openstack/openstack-ansible-os_tempest/blob/master/vars/debian.yml#L32-L36 to another pace, or I should change "owner" of the note? | 14:09 |
| noonedeadpunk | s/pace/place/ | 14:09 |
| odyssey4me | noonedeadpunk you can own the comment for any new files - I won't be chasing you about copyright for a comment ;) | 14:11 |
| *** jbadiapa has joined #openstack-ansible | 14:11 | |
| noonedeadpunk | tbh it's easier to just copy entire comment, but I was worring that you may chasing me that wasn't placing comment in that place:) | 14:12 |
| noonedeadpunk | btw, odyssey4me, are you going to be in Shanghai? | 14:13 |
| odyssey4me | noonedeadpunk I won't be in Shanghai, but I will be back at the next summit/ptg. I'm still trying to get to grips with everything in TripleO. :) | 14:15 |
| noonedeadpunk | Yeah, TripleO has really good team now, especially after you and cloudnull joined them | 14:17 |
| *** ppetit has joined #openstack-ansible | 14:18 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_sahara master: service_setup: refactor service setup to a single file https://review.opendev.org/675167 | 14:20 |
| openstackgerrit | Merged openstack/openstack-ansible-os_blazar master: Replace include with import https://review.opendev.org/675093 | 14:20 |
| openstackgerrit | Merged openstack/openstack-ansible-os_placement master: Convert dynamic includes to static imports https://review.opendev.org/675161 | 14:22 |
| openstackgerrit | Merged openstack/openstack-ansible-os_masakari master: service_setup: refactor service setup to a single file https://review.opendev.org/675141 | 14:23 |
| openstackgerrit | Merged openstack/openstack-ansible-os_congress master: Replace include with import https://review.opendev.org/675095 | 14:29 |
| *** dave-mccowan has quit IRC | 14:30 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_designate master: Convert dynamic includes to static imports https://review.opendev.org/675094 | 14:30 |
| openstackgerrit | Merged openstack/openstack-ansible-os_mistral master: service_setup: refactor service setup to a single file https://review.opendev.org/675143 | 14:32 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_tempest master: Add mixed setup for debian based OS https://review.opendev.org/675353 | 14:32 |
| openstackgerrit | Merged openstack/openstack-ansible-os_magnum master: service_setup: refactor service setup to a single file https://review.opendev.org/675100 | 14:33 |
| openstackgerrit | Merged openstack/openstack-ansible-os_octavia master: Convert dynamic includes to static imports https://review.opendev.org/675144 | 14:33 |
| openstackgerrit | Merged openstack/openstack-ansible-os_octavia master: service_setup: refactor service setup to a single file https://review.opendev.org/675153 | 14:33 |
| openstackgerrit | Merged openstack/openstack-ansible-os_panko master: service_setup: refactor service setup to a single file https://review.opendev.org/675159 | 14:43 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder master: Replace include with import https://review.opendev.org/675091 | 14:44 |
| *** vincelevi has joined #openstack-ansible | 14:45 | |
| vincelevi | hello, i'm trying to setup an openstack with a single admin node, I think I stumbled upon that bug : https://bugs.launchpad.net/openstack-ansible/+bug/1778914 | 14:46 |
| openstack | Launchpad bug 1778914 in openstack-ansible "galera server binds to 0.0.0.0 address" [Low,Confirmed] | 14:46 |
| openstackgerrit | Merged openstack/openstack-ansible-os_ironic master: service_setup: refactor service setup to a single file https://review.opendev.org/675090 | 14:47 |
| vincelevi | in comment #1 someone proposed 2 workarounds, I would like to use the "run without haproxy", but don't know how to do that, can someone push me in the right direction... | 14:48 |
| vincelevi | evrardjp: I think it was your suggestion in that comment | 14:49 |
| noonedeadpunk | vincelevi: I guess you may try commenting out haproxy setup playbook from setup-infrastructure playbook | 14:50 |
| noonedeadpunk | but there might be better solution | 14:51 |
| vincelevi | noonedeadpunk: thanks for the hint, I didn't expect it to be that simple :-) | 14:51 |
| openstackgerrit | Merged openstack/openstack-ansible-os_aodh master: service_setup: refactor service setup to a single file https://review.opendev.org/673495 | 14:54 |
| *** nurdie has joined #openstack-ansible | 14:54 | |
| noonedeadpunk | As I said there should be better solution, which just prevent inventory for creating record for haproxy group | 14:54 |
| noonedeadpunk | But I can't find this right now... | 14:55 |
| vincelevi | I was afraid it would be more complex than that because of the first note in https://docs.openstack.org/openstack-ansible-haproxy_server/latest/configure-haproxy.html | 14:55 |
| vincelevi | "For a successful installation, you require a load balancer." | 14:55 |
| noonedeadpunk | so since you have 1 copy of all api's you have nothing to balance... But yeah, there might be caveats I can't instantly recall | 14:57 |
| evrardjp | yeah you will still need a load balancer :) | 14:58 |
| evrardjp | vincelevi: you don't want to run with lxc containers? | 14:59 |
| evrardjp | it works just fine on a single node :) | 14:59 |
| noonedeadpunk | that is the easiest solution for sure ^ :) | 14:59 |
| vincelevi | That's my first use of openstack-ansible, so I'm discovering all of this (not new to OS admin, but it was 5 year ago) and yes I run the services in lxc containers | 15:00 |
| noonedeadpunk | So you shouldn't hit that for sure... | 15:00 |
| mnaser | noonedeadpunk: is https://review.opendev.org/#/c/675077/ the official fix for the ceph issue? | 15:01 |
| vincelevi | my setup would be a single admin node and a few dual services compute+cinder nodes | 15:01 |
| noonedeadpunk | mnaser: not yet. working on the better solution. It was just to find the root cause | 15:01 |
| mnaser | i see, so our gate is still broken for now? bummer :< | 15:02 |
| noonedeadpunk | yep:( | 15:02 |
| mnaser | i am a bit busy with some stuff.. but hopefully ill iron them out and maybe get some help if you havent gotten it working by then | 15:02 |
| noonedeadpunk | mnaser: actually, looks like it works since config is there https://logs.opendev.org/77/675077/4/check/openstack-ansible-deploy-aio_ceph-ubuntu-bionic/b2b581b/logs/etc/host/nova/nova.conf.txt.gz | 15:04 |
| noonedeadpunk | But I feel like I'm missing smth | 15:04 |
| vincelevi | something maybe related, is that I didn't fully understood internal_lb_vip_address / external_lb_vip_address and I put the ip of my admin node (on br-mgmt) in both, that is maybe wrong... | 15:04 |
| mnaser | noonedeadpunk: well technically we dont test a scenario where we use local storage + ceph for cinder | 15:04 |
| *** miloa has quit IRC | 15:04 | |
| mnaser | so its going to be impossible to test unless we do it against a dev cluster or something | 15:04 |
| noonedeadpunk | mnaser: I'm thinking about making this fact cacheable... | 15:05 |
| noonedeadpunk | so that it won't be out of scope (since it seems that it is for some reason)... | 15:06 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_heat stable/stein: Convert dynamic includes to static imports https://review.opendev.org/675377 | 15:08 |
| *** goldenfri has joined #openstack-ansible | 15:10 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_designate stable/stein: Convert dynamic includes to static imports https://review.opendev.org/675380 | 15:12 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_octavia stable/stein: Convert dynamic includes to static imports https://review.opendev.org/675381 | 15:14 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Fix condition of defining nova_cinder_rbd_inuse https://review.opendev.org/675077 | 15:16 |
| ppetit | Hi there! | 15:16 |
| ppetit | Couple quick questions if I may.... | 15:17 |
| noonedeadpunk | at least you may try :P | 15:17 |
| ppetit | It seems neutron LBaas V2 is not quite supported or encouraged. Is it correct ? | 15:17 |
| noonedeadpunk | ppetit: I gues it's replaced by octavia v2 | 15:18 |
| ppetit | Okay got it | 15:18 |
| ppetit | Now w.r.t documentation. | 15:18 |
| johnsom | ppetit Correct, neutron-lbaas has gone end-of-life. See: https://wiki.openstack.org/wiki/Neutron/LBaaS/Deprecation | 15:18 |
| *** jbadiapa has quit IRC | 15:18 | |
| johnsom | ppetit the LBaaS V2 API however lives on in Octavia | 15:19 |
| ppetit | Octavia is compatible with LBaas V2 API right. | 15:20 |
| johnsom | Correct, Octavia's API is a superset of the LBaaS v2 API | 15:20 |
| ppetit | Thank you! | 15:20 |
| ppetit | Then I noticed there use to be an OSA Install Guide until at least the Mikata release with a detailed content about how to install services ... | 15:21 |
| ppetit | Like in https://docs.openstack.org/openstack-ansible/mitaka/index.html | 15:22 |
| ppetit | The Configuration chapter for instance is well detailed. | 15:23 |
| ppetit | I do not seem the same level of details in the actual Deployment Guide unfortunately. | 15:23 |
| ppetit | Why has this good content bee dropped ? | 15:23 |
| *** jbadiapa has joined #openstack-ansible | 15:30 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_zun master: service_setup: refactor service setup to a single file https://review.opendev.org/675300 | 15:33 |
| *** spsurya has quit IRC | 15:34 | |
| *** jbadiapa has quit IRC | 15:35 | |
| noonedeadpunk | ppetit: isn't it what you are looking for? https://docs.openstack.org/project-deploy-guide/openstack-ansible/latest/ | 15:37 |
| ppetit | noonedeadpunk: Yes it is usually :) | 15:39 |
| noonedeadpunk | Ah, you mean detalisation | 15:40 |
| noonedeadpunk | So I guess the reason was that it's pretty hard to maintain documentation ,as it's get outdated. So it's much easier just to include defaults variable files with comments regarding some tricky variables | 15:42 |
| *** goldenfri has quit IRC | 15:42 | |
| *** gyee has joined #openstack-ansible | 15:43 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: docs: Release OSA Stein https://review.opendev.org/675385 | 15:47 |
| ppetit | Yes. I meant that to a certain extent that Install Guide had the merit to be fairly detailed about severals aspects of the config process while the actual doc is somewhat less explanatory. | 15:48 |
| ppetit | That being said I understand the burden of maintaining the doc up to date. | 15:50 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_tempest master: Add mixed setup for debian based OS https://review.opendev.org/675353 | 15:53 |
| *** the_intern has joined #openstack-ansible | 16:03 | |
| the_intern | Hey y'all, quick question. Where can I find the default haproxy settings for openstack services? Would they be somewhere in the haproxy-server role? | 16:10 |
| openstackgerrit | Merged openstack/openstack-ansible-os_tacker master: service_setup: refactor service setup to a single file https://review.opendev.org/675289 | 16:16 |
| jamesdenton | the_intern some are here: https://github.com/openstack/openstack-ansible/blob/master/inventory/group_vars/haproxy/haproxy.yml | 16:17 |
| jamesdenton | others in the role | 16:17 |
| jamesdenton | but there may be more i'm not aware of | 16:17 |
| the_intern | awesome, thanks jamesdenton | 16:18 |
| jamesdenton | is there something in particular you're looking for? | 16:18 |
| the_intern | trying to see if there are defaults for Manila on CephFS | 16:19 |
| openstackgerrit | Merged openstack/openstack-ansible-os_trove master: service_setup: refactor service setup to a single file https://review.opendev.org/675293 | 16:19 |
| noonedeadpunk | the_intern: Manila is hardly operatable I guess - it has some problems inside manila role itself | 16:19 |
| noonedeadpunk | Some of them has been just closed https://review.opendev.org/#/q/owner:noonedeadpunk%2540ya.ru+project:openstack/openstack-ansible-os_manila | 16:22 |
| noonedeadpunk | but yeah, there should be manila endpoint in haproxy | 16:23 |
| tbarron | and offhand that is nothing special https://github.com/openstack/openstack-ansible/blob/master/inventory/group_vars/haproxy/haproxy.yml#L268 | 16:24 |
| *** the_intern has quit IRC | 16:24 | |
| tbarron | and is for the api service so is completely independent of whether CephFS or other storage is used to back the manila-share service | 16:25 |
| noonedeadpunk | tbarron: hope it will pass tests soon.... Now we've discovered issue with nova-compute deployemnt which may ruin tepmest test for manial | 16:26 |
| tbarron | noonedeadpunk: yeah I saw that but | 16:26 |
| tbarron | noonedeadpunk: all that nova and cinder stuff is for the "generic" back end -- nova instances as "share servers" and cinder volumes with a filesystem as the backing store for shares | 16:27 |
| tbarron | noonedeadpunk: I don't know of anyone who uses that back end in production (but am willing to learn) | 16:28 |
| tbarron | noonedeadpunk: we don't have active maintainers for it in manila project itself any more | 16:28 |
| vincelevi | I fixed my haproxy problem, but now I got a disk full, the lxc containers are using the /var/lib fs despite the presence of "lxc" lvm VG (the OSA doc let me think that just creating the VG would be sufficient) what am I missing ? | 16:28 |
| noonedeadpunk | tbarron: so you mean that we may easily drop generic backend? | 16:28 |
| tbarron | noonedeadpunk: well it's kinda up to you and what the objectives of having manila support in OSA are | 16:29 |
| tbarron | noonedeadpunk: it's unfortunately still the default back end and is used in manila CI to do things with a software defined back end that otherwise | 16:30 |
| tbarron | noonedeadpunk: you need to buy a NetApp or DellEMC etc. for | 16:30 |
| noonedeadpunk | So I'm not actually manila expert, but I guess we are aiming to support it as production ready as well. | 16:30 |
| tbarron | noonedeadpunk: but in production to my knowledge people don't use it, if they want a softwared defined back end for manila | 16:31 |
| tbarron | noonedeadpunk: and many, maybe most do! | 16:31 |
| tbarron | noonedeadpunk: they use CephFS or GlusterFS | 16:31 |
| tbarron | noonedeadpunk: often with NFS gateway | 16:32 |
| *** allanb has joined #openstack-ansible | 16:32 | |
| allanb | OSA isn't tested with debian is it? just ubuntu and centos? | 16:33 |
| tbarron | noonedeadpunk: so it *might* be a justifiable stance for OSA to focus on a set of viable production back ends for manila and | 16:33 |
| admin0 | is anyone using LXD with ceph ? i added 2 lxd nodes on an all ceph env, but they just use local storage and not ceph ( i expected it would have used ceph ) | 16:33 |
| tbarron | noonedeadpunk: simplify what it needs to work with | 16:33 |
| allanb | i tried w/ some debian9 vm's and it failed on trying to install 'snap' packages, which i believe are only on ubuntu. could be wrong. | 16:33 |
| noonedeadpunk | tbarron: so the thing is, that right now we're failing NFS test as wel https://logs.opendev.org/33/660333/10/check/openstack-ansible-deploy-aio_metal-debian-stable/b6ceb19/logs/openstack/aio1-utility/stestr_results.html | 16:34 |
| noonedeadpunk | tbarron: And I guess manila still requires an instance made from service-image? | 16:35 |
| tbarron | noonedeadpunk: we have to distinguish between the back end and the protocol, that's NFS protocol but I bet you are using the "generic" (nova/cinder) back end | 16:35 |
| noonedeadpunk | even for nfs? | 16:35 |
| noonedeadpunk | oh... | 16:36 |
| tbarron | noonedeadpunk: if we look at the start of the manila-share log and at its manila.conf for which are the enabled_share_backends that will tell you | 16:36 |
| *** rpittau is now known as rpittau|afk | 16:36 | |
| tbarron | that generic back end can do either nfs or cifs share protocol | 16:36 |
| tbarron | but lots of other back ends do these too | 16:37 |
| tbarron | cephfs on the other hand does native cephfs or nfs (using a ganesha gateway) | 16:37 |
| tbarron | glusterfs does native glusterfs or nfs (with gateway) | 16:37 |
| tbarron | vendor storage typically does nfs and sometimes cifs | 16:38 |
| tbarron | and there are hdfs, maprfs, etc. but let's put those aside for now | 16:38 |
| noonedeadpunk | tbarron: hm, we've set nfs-share1 as a default type https://logs.opendev.org/33/660333/10/check/openstack-ansible-deploy-aio_metal-debian-stable/b6ceb19/logs/etc/host/manila/manila.conf.txt.gz | 16:38 |
| gouthamr | also, if using the generic driver (or the container driver, or NetApp/Dell-EMC/VendorXYZ with config opt driver_handles_share_servers=True), you will see cleanup errors in tempest if tempest's dynamic credentials are being used: auth.use_dynamic_credentials should be set to False. | 16:39 |
| tbarron | noonedeadpunk: look at enabled_share_backends and enabled_share_protocols | 16:40 |
| tbarron | noonedeadpunk: nfs_share1 is an LVM back end | 16:40 |
| tbarron | noonedeadpunk: reasonable choice for simplicity but probably not for production | 16:40 |
| vincelevi | OK looks like I have to set lxc_container_backing_store=lvm, where do I put that ? in openstack_user_config.yml ? as a top level variable override ? | 16:41 |
| *** chkumar|ruck is now known as raukadah | 16:42 | |
| tbarron | noonedeadpunk: you might want to consider only enabling LVM (disabling generic back end) for now and we can git it working end-to-end first | 16:43 |
| tbarron | noonedeadpunk: then enable more back ends if that is useful | 16:43 |
| tbarron | get it :) | 16:43 |
| noonedeadpunk | tbarron: ok, thank you for your input, will try to make smth out of this discussion. | 16:43 |
| noonedeadpunk | so users will be free to add whatever backend they want. For testing I think LVM and CephFS should be enough all in all. But yeah, I'll probably start from LVM as an easiest thng to test on | 16:45 |
| tbarron | noonedeadpunk: gouthamr did some nice docs here that include sample local.conf etc. for these back ends: | 16:47 |
| tbarron | https://docs.openstack.org/manila/latest/contributor/development-environment-devstack.html#sample-local-conf-files-that-get-you-started | 16:47 |
| *** tosky has quit IRC | 16:49 | |
| noonedeadpunk | great, thanks! | 16:50 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Fix condition of defining nova_cinder_rbd_inuse https://review.opendev.org/675077 | 16:53 |
| tbarron | noonedeadpunk: Thank you for working on this! | 16:58 |
| *** the_intern has joined #openstack-ansible | 16:59 | |
| *** BjoernT has quit IRC | 17:02 | |
| *** BjoernT has joined #openstack-ansible | 17:03 | |
| *** BjoernT has quit IRC | 17:03 | |
| *** BjoernT_ has joined #openstack-ansible | 17:04 | |
| *** BjoernT_ has quit IRC | 17:04 | |
| *** BjoernT has joined #openstack-ansible | 17:05 | |
| *** BjoernT has quit IRC | 17:05 | |
| *** BjoernT has joined #openstack-ansible | 17:07 | |
| *** BjoernT has quit IRC | 17:07 | |
| *** BjoernT_ has joined #openstack-ansible | 17:08 | |
| *** BjoernT_ has quit IRC | 17:08 | |
| *** udesale has quit IRC | 17:08 | |
| *** BjoernT has joined #openstack-ansible | 17:08 | |
| *** BjoernT has quit IRC | 17:09 | |
| *** BjoernT has joined #openstack-ansible | 17:09 | |
| *** BjoernT has quit IRC | 17:10 | |
| *** BjoernT_ has joined #openstack-ansible | 17:10 | |
| *** BjoernT_ has quit IRC | 17:10 | |
| *** BjoernT has joined #openstack-ansible | 17:11 | |
| *** BjoernT has quit IRC | 17:11 | |
| *** BjoernT has joined #openstack-ansible | 17:11 | |
| *** BjoernT has quit IRC | 17:12 | |
| *** markvoelker has quit IRC | 17:12 | |
| *** BjoernT has joined #openstack-ansible | 17:13 | |
| *** BjoernT has quit IRC | 17:14 | |
| *** BjoernT_ has joined #openstack-ansible | 17:14 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_octavia stable/stein: Convert dynamic includes to static imports https://review.opendev.org/675381 | 17:14 |
| *** BjoernT_ has quit IRC | 17:14 | |
| *** BjoernT has joined #openstack-ansible | 17:15 | |
| *** BjoernT has joined #openstack-ansible | 17:15 | |
| *** BjoernT has quit IRC | 17:16 | |
| *** BjoernT has joined #openstack-ansible | 17:16 | |
| *** BjoernT has quit IRC | 17:17 | |
| *** BjoernT_ has joined #openstack-ansible | 17:17 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_barbican master: service_setup: refactor service setup to a single file https://review.opendev.org/673491 | 17:17 |
| *** BjoernT_ has quit IRC | 17:17 | |
| *** BjoernT has joined #openstack-ansible | 17:18 | |
| *** BjoernT has quit IRC | 17:18 | |
| *** BjoernT has joined #openstack-ansible | 17:19 | |
| logan- | noonedeadpunk: would love to learn more about whats prompting https://review.opendev.org/#/c/675077 if you can point me to any bugs/logs. mostly just curious but happy to help if you need any assistance with it. | 17:19 |
| *** BjoernT has quit IRC | 17:20 | |
| *** BjoernT_ has joined #openstack-ansible | 17:20 | |
| noonedeadpunk | logan-: we've got broken gates right now | 17:20 |
| noonedeadpunk | https://review.opendev.org/#/c/674653/ | 17:20 |
| *** BjoernT_ has quit IRC | 17:20 | |
| *** BjoernT has joined #openstack-ansible | 17:21 | |
| logan- | thanks | 17:21 |
| *** BjoernT has quit IRC | 17:21 | |
| noonedeadpunk | and patchset 4 was passing so now I'm trying to play around. | 17:22 |
| *** BjoernT has joined #openstack-ansible | 17:22 | |
| noonedeadpunk | So if you have any ideas why this happens this will be highly appreciated | 17:22 |
| *** BjoernT has quit IRC | 17:22 | |
| *** BjoernT has joined #openstack-ansible | 17:22 | |
| *** BjoernT has quit IRC | 17:23 | |
| *** BjoernT_ has joined #openstack-ansible | 17:23 | |
| *** vincelevi has quit IRC | 17:23 | |
| logan- | skipping the set fact is so weird. behavior change with ansible 2.8 maybe causing it maybe? | 17:24 |
| noonedeadpunk | Actually that happened right after you fixed variable in nova:) | 17:25 |
| noonedeadpunk | So I don't know if it's ansible 2.8 or we were just living somehow with it | 17:25 |
| noonedeadpunk | I mean https://opendev.org/openstack/openstack-ansible-os_nova/commit/e57db22374439d7d6602659092acee27bc72fd23 | 17:25 |
| *** markvoelker has joined #openstack-ansible | 17:26 | |
| logan- | well i mention that because i do know we've gone thru several iterations over the years with that var due to ansible fact handling changes, so thats the first thing i think of when i see weird issues with it | 17:26 |
| logan- | maybe we should default nova_cinder_rbd_inuse in the nova role to false | 17:27 |
| *** BjoernT_ has quit IRC | 17:28 | |
| logan- | and then check the gate logs for the ceph job to make sure it correctly goes to true and handles building nova.conf properly for the True state | 17:28 |
| *** ppetit has quit IRC | 17:30 | |
| noonedeadpunk | the point is that we don't test mixed thing - when nova do not use ceph, while cinder does | 17:30 |
| *** kopecmartin is now known as kopecmartin|off | 17:31 | |
| noonedeadpunk | so we won't know how it works in such cases which was the point I guess | 17:31 |
| noonedeadpunk | but yeah. strange that it doesn't set_fact... | 17:32 |
| noonedeadpunk | I mean it skips it | 17:32 |
| noonedeadpunk | oh... lxc is not in nova_compute | 17:33 |
| noonedeadpunk | so we won't have fact on this stage.... | 17:34 |
| logan- | gotcha | 17:39 |
| logan- | well worst case we could do a DNM patch changing our ceph job to not configure the libvirt pool just to make sure it works | 17:39 |
| logan- | idk if we should add a permanent job for that, it might not be a bad idea, but its pretty easily testable as a one-off in the gate either way | 17:40 |
| noonedeadpunk | so probably we can just simplify condition to "inventory_hostname == ((groups['nova_all'] | intersect(ansible_play_hosts)) | list)[0]" | 17:41 |
| logan- | oh i see the host it is failing on is not even the compute, its the api container | 17:41 |
| noonedeadpunk | yep | 17:41 |
| logan- | the fact is not even needed there, so the default would satisfy that just fine | 17:42 |
| noonedeadpunk | since api s configured first in terms of lxc | 17:42 |
| logan- | i think its probably just a situation where we're missing a role default that needs to be set | 17:42 |
| noonedeadpunk | It's not missing | 17:43 |
| noonedeadpunk | https://opendev.org/openstack/openstack-ansible-os_nova/src/branch/master/defaults/main.yml#L394 | 17:43 |
| noonedeadpunk | That's what frustrated me first:) | 17:43 |
| logan- | gotcha | 17:43 |
| *** ianychoi has quit IRC | 17:43 | |
| logan- | then we go and pass in that un-referencable fact in the playbook | 17:44 |
| noonedeadpunk | So I'd say that it's safe to define fact just at nova_all | 17:44 |
| noonedeadpunk | yep | 17:44 |
| logan- | now that i get the picture I think https://review.opendev.org/#/c/675077/4/playbooks/common-playbooks/nova.yml is a reasonable fix | 17:45 |
| logan- | you may be able to default(omit) there | 17:45 |
| logan- | which would be even better | 17:45 |
| logan- | then we can defer to the role default | 17:45 |
| logan- | otherwise default(False) there is probably next best | 17:46 |
| *** ianychoi has joined #openstack-ansible | 17:47 | |
| noonedeadpunk | so, why don't you like simplifying condition when fact is gathered? If take into account that nova.conf is going to be devided into nova.conf and nova-compute.cnf? | 17:49 |
| noonedeadpunk | So we'll have fact when role wll be runing against the first nova host/container | 17:50 |
| logan- | if we don't need to set a fact to override role default behavior, let's not set it. api deploys properly with the role default, so we should use it imo. | 17:52 |
| logan- | and nova.conf is not divided by this fact, there are plenty of group-specific blocks deployed in many of our configs... especially nova. its not a new thing at all :) | 17:54 |
| noonedeadpunk | I mean we're unlikely have deployment with nova-api but without nova-compute... So fact will be set sooner or later. And since we're setting it on the deployemnt host, it will be kinda global for the rest of the services | 17:54 |
| *** ianychoi_ has joined #openstack-ansible | 17:54 | |
| *** ianychoi_ has quit IRC | 17:55 | |
| noonedeadpunk | logan-: so nova has to be splited since definition of api_database section makes compute try to reach database with it own | 17:56 |
| noonedeadpunk | So on aio_metal compute should have big problems | 17:57 |
| noonedeadpunk | https://review.opendev.org/#/c/675148/ | 17:57 |
| *** BjoernT has joined #openstack-ansible | 17:57 | |
| *** ianychoi has quit IRC | 17:58 | |
| logan- | yep.. im not sure what you mean by "splitted" though. it sounds like we have a problem with the database config block conditional when deploying on metal, a totally separate issue that will require a fix to the conditional in the nova.conf template. | 17:58 |
| *** zirpu has quit IRC | 17:59 | |
| logan- | i don't follow how it relates to the rbd fact fix? | 17:59 |
| noonedeadpunk | logan-: yeah, maybe you're right and it don't... I just mean that condition to gather fact only on compute the frst compute node doesn't make much sense to me, since it will be gathered anyway | 18:00 |
| logan- | it is a compute specific var override.. we don't need to override the role default otherwise | 18:01 |
| logan- | there's no need to run that set fact for other groups | 18:01 |
| logan- | looking at the playbook i think it should also work right in a metal scenario where api and compute are collocated right? | 18:02 |
| noonedeadpunk | So the only place where it's used is a part of nova_rbd_inuse which is vitalhere https://opendev.org/openstack/openstack-ansible-os_nova/src/branch/master/templates/nova.conf.j2#L253 | 18:03 |
| openstackgerrit | Logan V proposed openstack/openstack-ansible master: Omit nova_cinder_rbd_inuse when it is not set https://review.opendev.org/675429 | 18:04 |
| noonedeadpunk | so that how I came to config file separation - this part is supposed to be only in nova-compute.conf but not in nova.conf I guess. | 18:04 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Set default for nova_cinder_rbd_inuse https://review.opendev.org/675077 | 18:04 |
| noonedeadpunk | logan-: oh, you've placed yours first.... | 18:05 |
| logan- | no worries, just wanted to get it pushed to see the gate run | 18:05 |
| noonedeadpunk | cool | 18:05 |
| jrosser | logan-: you sure you can use omit like that? Isn’t it strictly for module parameters only? | 18:05 |
| logan- | i'll abandon mine and we can watch your patch | 18:05 |
| logan- | jrosser: iirc it is valid in this context | 18:06 |
| noonedeadpunk | logan-: I've already abandoned mine:) | 18:06 |
| logan- | oh ok | 18:06 |
| openstackgerrit | Logan V proposed openstack/openstack-ansible master: [DNM] Test ceph without defining nova libvirt pool https://review.opendev.org/675430 | 18:08 |
| noonedeadpunk | Sorry need to run away for some time | 18:11 |
| logan- | np will keep an eye on the patches and see how it goes | 18:11 |
| noonedeadpunk | Hope be back until ci results:) | 18:12 |
| *** nurdie_ has joined #openstack-ansible | 18:59 | |
| *** nurdie has quit IRC | 19:03 | |
| *** BjoernT has quit IRC | 19:15 | |
| *** BjoernT has joined #openstack-ansible | 19:16 | |
| allanb | the base container image for debian9 (stretch) seems to be missing apt-listchanges, though it has the config for apt in /etc/apt/apt.conf.d/20listchanges | 19:18 |
| allanb | i'll open a bug for that. | 19:18 |
| noonedeadpunk | logan-: so seems that omit work as expected | 19:18 |
| noonedeadpunk | and since it's really cleaner than default(False), see no reason not to place it | 19:19 |
| *** BjoernT has quit IRC | 19:20 | |
| *** nurdie_ has quit IRC | 19:25 | |
| *** zirpu has joined #openstack-ansible | 19:25 | |
| *** ivve has quit IRC | 19:29 | |
| noonedeadpunk | allanb: just post bug here once placed | 19:32 |
| allanb | https://bugs.launchpad.net/openstack-ansible/+bug/1839535 | 19:39 |
| openstack | Launchpad bug 1839535 in openstack-ansible "debian9 container image is missing apt-listchanges package, but has the apt.conf.d/20listchanges config file, causing failure." [Undecided,New] | 19:39 |
| *** noonedeadpunk has quit IRC | 19:42 | |
| *** noonedeadpunk has joined #openstack-ansible | 19:46 | |
| noonedeadpunk | test | 19:46 |
| *** kplant has quit IRC | 19:46 | |
| noonedeadpunk | so this looks again like variables scope issues caused by my patch https://logs.opendev.org/53/675353/2/check/tripleo-ci-centos-7-standalone-os-tempest/dca8eec/job-output.txt.gz#_2019-08-08_17_07_08_928815 | 19:47 |
| noonedeadpunk | But the thing is that only triple-o check fails with that | 19:47 |
| noonedeadpunk | And variable is defined in vars https://review.opendev.org/#/c/675353/2/vars/main.yml | 19:49 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_tempest master: Add mixed setup for debian based OS https://review.opendev.org/675353 | 19:54 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_tempest master: Add mixed setup for debian based OS https://review.opendev.org/675353 | 19:58 |
| *** jbadiapa has joined #openstack-ansible | 20:00 | |
| *** noonedeadpunk has quit IRC | 20:03 | |
| *** the_intern has quit IRC | 20:08 | |
| *** jbadiapa has quit IRC | 20:12 | |
| donnyd | osa still uses lxc correct? | 20:18 |
| allanb | yes | 20:21 |
| *** ansmith has quit IRC | 20:22 | |
| jrosser | donnyd: youve got options for lxc or bare metal | 20:23 |
| donnyd | thanks jrosser :) | 20:24 |
| *** ivve has joined #openstack-ansible | 20:32 | |
| rholloway | is there a proper way to shut down a cluster that was deployed with osa? I see in the guide a few things for specific services like galera / rmq (essentially last node down should be first up), but curious if there is anything documented around full cluster in cases needed | 20:37 |
| rholloway | I kinda broke my first osa deployment yesterday I think rebooting all the nodes and not being careful enough about it or tracking order for things like galera, I probably need a more formal process around that | 20:38 |
| spotz | rholloway: I know for an AIO there were instructions for reboooting which might cover the steps you're looking for | 20:51 |
| admin0 | donnyd, i setup lxc just like 2 days ago .. everything works except console, and it is using local storage instead of ceph | 20:52 |
| donnyd | you mean for the control plane or for nova? | 20:53 |
| admin0 | nova | 20:53 |
| donnyd | oh nice | 20:53 |
| admin0 | well, control plane automatically uses that | 20:53 |
| spotz | admin0: Did you see my reply to you the other day on the ops channel? | 20:53 |
| admin0 | yep .. i saw it the other day though .. i said good night here and was in bed | 20:54 |
| spotz | admin0: Yeah I'd seen the good night part and then the question there after:) | 20:54 |
| admin0 | its working nicely now | 20:54 |
| *** ansmith has joined #openstack-ansible | 20:56 | |
| rholloway | hmm only thing I see about AIO reboot is https://docs.openstack.org/openstack-ansible/stein/user/aio/quickstart.html and just has a quick note about galera, maybe galera is the main thing to worry about and I should just be sure I do things in right order | 20:56 |
| rholloway | using multiple physical nodes, not AIO | 20:57 |
| spotz | rholloway: Yeah that's the only docs I could think of. Maybe one of the others will pop in | 20:58 |
| openstackgerrit | Douglas Mendizábal proposed openstack/ansible-role-thales-hsm stable/stein: Ensure libnsl is installed https://review.opendev.org/675464 | 21:04 |
| openstackgerrit | Douglas Mendizábal proposed openstack/ansible-role-thales-hsm stable/stein: Fix typo in client playbook https://review.opendev.org/675465 | 21:07 |
| *** markvoelker has quit IRC | 21:36 | |
| *** markvoelker has joined #openstack-ansible | 21:53 | |
| *** markvoelker has quit IRC | 21:57 | |
| *** jsquare has quit IRC | 22:07 | |
| *** jsquare has joined #openstack-ansible | 22:07 | |
| *** jsquare has quit IRC | 22:08 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!