| *** cshen has joined #openstack-ansible | 00:12 | |
| *** cshen has quit IRC | 00:17 | |
| *** gyee has quit IRC | 01:23 | |
| *** stingrayza has quit IRC | 01:51 | |
| *** stingrayza has joined #openstack-ansible | 01:53 | |
| *** jcath has joined #openstack-ansible | 01:57 | |
| jcath | hello, friends. is it possible to limit containers' ip address range every specific host? let's say I want host 172.16.0.1 with containers' ip range 172.16.0.10-172.16.0.30 | 01:59 |
|---|---|---|
| *** cshen has joined #openstack-ansible | 02:13 | |
| *** cshen has quit IRC | 02:17 | |
| *** markvoelker has joined #openstack-ansible | 03:06 | |
| *** markvoelker has quit IRC | 03:11 | |
| *** cshen has joined #openstack-ansible | 03:15 | |
| *** cshen has quit IRC | 03:19 | |
| *** jcath has quit IRC | 03:26 | |
| *** jcath has joined #openstack-ansible | 03:49 | |
| *** rh-jelabarre has quit IRC | 04:16 | |
| *** evrardjp has quit IRC | 04:33 | |
| *** evrardjp has joined #openstack-ansible | 04:33 | |
| *** udesale has joined #openstack-ansible | 04:41 | |
| *** markvoelker has joined #openstack-ansible | 04:44 | |
| *** markvoelker has quit IRC | 04:49 | |
| *** markvoelker has joined #openstack-ansible | 05:04 | |
| *** markvoelker has quit IRC | 05:09 | |
| *** cshen has joined #openstack-ansible | 05:15 | |
| *** cshen has quit IRC | 05:19 | |
| jrosser | jcath: can you explain why? each host in a different L3 subnet perhaps? | 05:34 |
| jcath | jrosser: i have 3 hosts, as I said the situation ,both nova-api and nova-compute on the same host, it does not work, with same nova.conf. so I put nova-api in a lxc container, and retain other on metal. but for easy management, I want the container take an ip just follow the physical hosts, (i.e. hosts is 172.16.0.1,.2,.3, I want the container take 172.16.0.4) | 05:38 |
| jrosser | ok, i'm not sure there is a way to do that | 05:39 |
| jrosser | the dynamic inventory generates the container ip addresses at random from the range you give in openstack_user_config | 05:41 |
| *** cshen has joined #openstack-ansible | 05:41 | |
| jcath | yes, i understand. may I change the address after apply setup-hosts.yml? in openstack_hostnames_ips.yml | 05:42 |
| jrosser | the container is already created at that point so the IP has get used in all sorts of places | 05:43 |
| jrosser | it might work, or you might need to unpick a bunch of stuff | 05:43 |
| jrosser | it is possible to define the inventory entirely manually also | 05:43 |
| jrosser | though i'm not sure thats a very well trodden path | 05:43 |
| jrosser | jcath: https://evrard.me/custom-inventories-with-openstack-ansible/ | 05:45 |
| jcath | thanks:) | 05:45 |
| *** cshen has quit IRC | 05:46 | |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-plugins master: Remove physical_host_addrs parameter from ssh plugin https://review.opendev.org/737220 | 05:50 |
| jrosser | jcath: if you really want nova compute and nova api on the same host it would be better to work on a patch that creates two config files | 05:51 |
| *** mgariepy has quit IRC | 05:52 | |
| jcath | yeah, i agree. just now I ran another round for testing, I definitely set the nova-api with is_metal, but the system still generate a lxc container ... maybe I should remove openstack_hostnames_ips.yml & openstack_inventory.json, then rerun the playbooks | 05:54 |
| *** mgariepy has joined #openstack-ansible | 06:10 | |
| *** pcaruana has joined #openstack-ansible | 06:20 | |
| *** mgariepy has quit IRC | 06:22 | |
| *** mgariepy has joined #openstack-ansible | 06:24 | |
| *** rpittau|afk is now known as rpittau | 06:34 | |
| *** miloa has joined #openstack-ansible | 06:39 | |
| *** cshen has joined #openstack-ansible | 06:59 | |
| *** markvoelker has joined #openstack-ansible | 07:05 | |
| *** jbadiapa has joined #openstack-ansible | 07:06 | |
| *** markvoelker has quit IRC | 07:11 | |
| *** tosky has joined #openstack-ansible | 07:35 | |
| *** markvoelker has joined #openstack-ansible | 07:49 | |
| *** markvoelker has quit IRC | 07:54 | |
| *** dpaclt has joined #openstack-ansible | 07:56 | |
| dpaclt | Hello All , i created a trove instance and its healthy and can telnet to it ... but the healthy status never gets completed ... Its stable/ussuri | 07:56 |
| jrosser | dpaclt: ussuri is still a release candidate so you are doing serious beta testing there | 08:08 |
| jrosser | and i think there is some network setup possibly needed to make trove work, br-dbaas and maybe a provider network | 08:08 |
| dpaclt | have created br-dbaas and a provider network too ... i think better to move to stable/train then | 08:30 |
| jrosser | if everything is otherwise working then you could probably stick where you are you are to debug it | 08:33 |
| dpaclt | Db instance is also working i can telnet to it without any issues.. only hiccup is the healthy status never get completed | 08:34 |
| jrosser | right - so something is wrong with the trove service getting that status | 08:35 |
| dpaclt | oh ok | 08:48 |
| jrosser | as far as i know trove communicates with the db instances using rabbitmq over the dbaas network | 08:49 |
| jrosser | but i've never deployed trove, so i can't give you specific debug tips | 08:49 |
| dpaclt | oh ok np ..Thanks jrosser | 08:57 |
| *** owalsh has quit IRC | 09:08 | |
| *** janno has quit IRC | 09:08 | |
| *** owalsh_ has joined #openstack-ansible | 09:08 | |
| *** janno has joined #openstack-ansible | 09:10 | |
| *** sshnaidm|afk is now known as sshnaidm|ruck | 09:11 | |
| jrosser | noonedeadpunk: i just run https://review.opendev.org/#/c/737220 in a multinode lab environment - same as the AIO it gives 2x speedup for re-running playbooks/lxc-container-create.yml | 09:19 |
| noonedeadpunk | we just need to make it working then... in terms of CI stuff | 09:21 |
| noonedeadpunk | yeah, the way we iterate over inventory can really slow down execution... | 09:22 |
| jrosser | tbh i am really not understanding the test case that is failing | 09:24 |
| jrosser | it's all to do with this https://github.com/openstack/openstack-ansible-plugins/blob/master/tests/inventory#L8-L12 | 09:25 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-openstack_hosts master: Fix CentOS functional test https://review.opendev.org/737331 | 09:25 |
| noonedeadpunk | having group twice is really weird | 09:34 |
| noonedeadpunk | however bug referenced there has some sence, but not sure it's valid nowadays | 09:34 |
| jrosser | the thing it fails on is deleting the non-existing container before creating it, i think https://zuul.opendev.org/t/openstack/build/b385be87635747b4bbfde5e72b67da7d/log/job-output.txt#2959 | 09:42 |
| jrosser | and i think it's trying to connect directly rather than with localhost | 09:43 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/ansible-role-systemd_networkd master: Add CentOS 8 support https://review.opendev.org/737122 | 09:44 |
| *** udesale_ has joined #openstack-ansible | 09:46 | |
| noonedeadpunk | wondering what pha var reflects... Like we don't return it or use in any way in code... But it somehow affects execution process (strategy) | 09:48 |
| noonedeadpunk | I mean the dropepd one from https://review.opendev.org/#/c/737220/9/strategy/linear.py | 09:49 |
| *** udesale has quit IRC | 09:49 | |
| noonedeadpunk | I guess it's used by strategy, as we're just overriding _queue_task... | 09:49 |
| *** markvoelker has joined #openstack-ansible | 09:50 | |
| noonedeadpunk | but can't find anywhere in https://github.com/ansible/ansible/blob/v2.9.9/lib/ansible/plugins/strategy/__init__.py | 09:51 |
| noonedeadpunk | so the idea was to connect always by IP if I got the idea. | 09:55 |
| *** markvoelker has quit IRC | 09:55 | |
| noonedeadpunk | and we reverting back to usage of hostnames? | 09:57 |
| *** jcath has quit IRC | 09:58 | |
| *** rohit02 has joined #openstack-ansible | 10:04 | |
| *** ut2k3 has joined #openstack-ansible | 10:07 | |
| ut2k3 | Hi guys. Whats the best-practice to retry "deploy-config-changes.yml" when some hosts failed on the run? There is no .retry file as well running the command again skips all stuff. | 10:07 |
| ut2k3 | Background is: I am about to do a major upgrade. | 10:09 |
| *** cshen has quit IRC | 10:10 | |
| *** dpaclt has quit IRC | 10:11 | |
| *** rpittau is now known as rpittau|bbl | 10:12 | |
| ut2k3 | Ok never mind... had too less coffee. | 10:15 |
| rohit02 | Hi team,openstack-ansible ussuri can someone share the docs/steps to integrate external ceph | 10:28 |
| *** jhesketh has quit IRC | 10:36 | |
| *** jhesketh has joined #openstack-ansible | 10:37 | |
| admin0 | ut2k3, u can use limits to run playbooks only for that hosts which fail | 10:41 |
| ut2k3 | Thanks. Just a last quick question. How can I add a host to the lxc_hosts group? I found only a away to remove with inventory manage but not to add. Background is that one host that should be inside that group is missing there. | 10:50 |
| *** cshen has joined #openstack-ansible | 10:56 | |
| jrosser | noonedeadpunk: yes you are right it is turning it to an IP with the old code http://paste.openstack.org/show/795092/ | 11:24 |
| jrosser | that is with my changes (top), without (bottom) | 11:24 |
| jrosser | the alternative is we look more at the current code to find exactly which bit is slow | 11:27 |
| noonedeadpunk | jrosser: don't see much difference. and do you have `infra1` defined in /etc/hosts file? | 11:32 |
| jrosser | well it's just the extra two lines of debug | 11:33 |
| jrosser | in this case infra1 resolves in DNS | 11:33 |
| jrosser | but thats i guess quite unusual | 11:34 |
| noonedeadpunk | what bothers me, is how do we connect to physical_host? so yeah, if we don't have it in dns (which is lamost always true I guess?) and not in hosts file, won't we end up in being unable to connect... | 11:35 |
| * noonedeadpunk needs to play with it... | 11:35 | |
| noonedeadpunk | porbably ansible does resolve things even without our overrides... | 11:35 |
| noonedeadpunk | btw was trying to reproduce https://34e9586208b3d2d1b836-40001c60cc2cc6a7bddcc7547f0d5486.ssl.cf2.rackcdn.com/689629/69/check/openstack-ansible-deploy-aio_distro_metal-centos-8/510e666/logs/ara-report/result/51d45044-905a-49b5-bdd1-015a247b9f35/ but I can't((( | 11:36 |
| *** jawad_axd has joined #openstack-ansible | 11:43 | |
| *** markvoelker has joined #openstack-ansible | 11:52 | |
| *** markvoelker has quit IRC | 11:57 | |
| *** rh-jelabarre has joined #openstack-ansible | 12:04 | |
| *** rpittau|bbl is now known as rpittau | 12:14 | |
| *** cshen has quit IRC | 12:32 | |
| openstackgerrit | Merged openstack/openstack-ansible-ops master: Stop to use the __future__ module. https://review.opendev.org/732882 | 12:37 |
| *** cshen has joined #openstack-ansible | 13:08 | |
| *** cshen has quit IRC | 13:13 | |
| *** cshen has joined #openstack-ansible | 13:18 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-openstack_hosts master: Fix CentOS functional test https://review.opendev.org/737331 | 13:37 |
| *** ut2k3 has quit IRC | 13:45 | |
| *** ut2k3 has joined #openstack-ansible | 13:47 | |
| *** spatel has joined #openstack-ansible | 13:51 | |
| *** sshnaidm|ruck is now known as sshnaidm|afk | 13:51 | |
| *** markvoelker has joined #openstack-ansible | 13:53 | |
| *** rohit02 has quit IRC | 13:55 | |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-plugins master: Optimise linear strategy access to physical host variables https://review.opendev.org/737543 | 13:56 |
| *** markvoelker has quit IRC | 13:57 | |
| *** spatel has quit IRC | 14:05 | |
| *** aedc_ has quit IRC | 14:10 | |
| *** ut2k3 has quit IRC | 14:11 | |
| *** ut2k3 has joined #openstack-ansible | 14:11 | |
| *** ut2k3 has quit IRC | 14:12 | |
| janno | when running "openstack-ansible os-nova-install.yml" the venv always gets rebuild because requirements.txt changes between compute hosts and api containers in Train because of nova_novnc_pip_packages | 14:12 |
| janno | is this a known issue? | 14:12 |
| *** sshnaidm|afk is now known as sshnaidm|ruck | 14:39 | |
| *** markvoelker has joined #openstack-ansible | 14:39 | |
| openstackgerrit | Merged openstack/openstack-ansible-memcached_server stable/stein: Reduce verbosity level when debug is true https://review.opendev.org/736181 | 14:41 |
| noonedeadpunk | janno: I guess it's not... | 14:42 |
| noonedeadpunk | jrosser: btw, you said you used roles outside of osa. but how did you handle config_template plugin? I don't see really handy way to get it installed, which is pretty upsetting | 14:43 |
| jrosser | let me look | 14:43 |
| *** markvoelker has quit IRC | 14:44 | |
| *** jawad_axd has quit IRC | 14:51 | |
| jrosser | noonedeadpunk: it looks like we just copied it wholesale into our repo with playbooks in | 14:57 |
| jrosser | which is sad | 14:57 |
| noonedeadpunk | yeah:( | 14:58 |
| noonedeadpunk | I'm not sure that action plugin can be shipped as collection.... | 14:58 |
| *** cshen has quit IRC | 14:59 | |
| jrosser | "Collections are a distribution format for Ansible content that can include playbooks, roles, modules, and plugins" ? | 15:02 |
| noonedeadpunk | then it's probably top 1 candidate to become a collection... but not sure how we will migrate... and we need CI spec at first I think | 15:06 |
| *** cshen has joined #openstack-ansible | 15:06 | |
| *** owalsh_ is now known as owalsh | 15:07 | |
| noonedeadpunk | yeah, you're absolutely right https://docs.ansible.com/ansible/devel/dev_guide/developing_collections.html#plugins-directory | 15:10 |
| *** cshen has quit IRC | 15:11 | |
| jrosser | ideally we could use the same stuff the openstack modules collection is using | 15:13 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-plugins master: Optimise linear strategy access to physical host variables https://review.opendev.org/737543 | 15:23 |
| *** spatel has joined #openstack-ansible | 15:24 | |
| jrosser | noonedeadpunk: ^ i narrowed it down to just one line of code causing the slowing down | 15:24 |
| noonedeadpunk | wow | 15:24 |
| spatel | is there an easy way to remove bulk host from aggregate host group? openstack aggregate remove host foo comp1,comp2,comp3 something like that? | 15:25 |
| jrosser | the one i've taken out returns *ALL* of the vars for every host | 15:25 |
| noonedeadpunk | Yes, I was just looking through it and was about to ask if that's the whle problem | 15:25 |
| jrosser | but still tests fail, i've just taken it off the stack of my other patches to test it in isolation | 15:25 |
| noonedeadpunk | yeah, no sense in getting host variables among all variables when we iterate over host... | 15:26 |
| jrosser | it seems to do the right thing, when i run here with -vvvv i still see it determining the IP for the physical host, so thats something at least | 15:27 |
| spatel | noonedeadpunk: jrosser how are you folks? let me know if you need my help in any area for testing etc.. | 15:29 |
| jrosser | tbh having to catch up on day job for a bit | 15:29 |
| jrosser | solid OSA stuff for a couple of weeks but i'm behind on other stuff now | 15:30 |
| spatel | I am also got little busy on my job stuff but soon will get free | 15:31 |
| * noonedeadpunk wondring if CI works at all for the repo | 15:31 | |
| noonedeadpunk | spatel: we got centos8 source metal jobs passing. | 15:32 |
| noonedeadpunk | I'm going to investigate what's wrong with distro ones | 15:32 |
| noonedeadpunk | and I guess lxc is also pretty close | 15:32 |
| spatel | phew!!! that was big win.. because i was seeing it failing on each checkout | 15:32 |
| spatel | I am waiting for centOS8 to get out soon :) | 15:33 |
| jrosser | the focal patch seems to refuse to merge | 15:33 |
| noonedeadpunk | yeah( | 15:33 |
| spatel | there must be some code cleaning up we can deal with them on our way right? | 15:33 |
| jrosser | trouble is a lot of the cleaning up is centos-7 related, sort of residual special cases for lxc2 and stuff | 15:34 |
| spatel | In August going to build new openstack cloud (100 nodes) so looking forward to deploy on centos8 | 15:34 |
| noonedeadpunk | btw we can join it with suse cleanup stuff | 15:34 |
| noonedeadpunk | but i'd first backport all and finall release U first | 15:35 |
| jrosser | once we merge/backport centos-8 to ussuri we can start cleanup on master | 15:35 |
| spatel | If we have stable centOS8 then i am sure people will go with CentOS8 instead using centOS7 | 15:35 |
| jrosser | we need a release that supports both | 15:35 |
| noonedeadpunk | (and merge https://review.opendev.org/737543 or smth regarding the topic) | 15:35 |
| jrosser | otherwise upgrades are horrific | 15:35 |
| spatel | +1 | 15:35 |
| jrosser | yeah i was hoping this linear strategy thing will help in CI jobs | 15:36 |
| jrosser | but it doesnt seem to have so bit an effect, but it's really really hard to tell | 15:37 |
| noonedeadpunk | kinda if you see 50% speedup in sandbox, that's pretty good result to try out | 15:38 |
| noonedeadpunk | as eventually in prod envs roles are really slow | 15:38 |
| noonedeadpunk | and this looks smth we can get profit pretty fast | 15:38 |
| jrosser | yeah it's absolutely needed for prod stuff | 15:39 |
| jrosser | point upgrades will get a significant boost with lots of "no change" tasks | 15:40 |
| *** cshen has joined #openstack-ansible | 15:40 | |
| jrosser | i wonder if this single thing as been so huge that it has made any other attempt we have done to optimise be hidden | 15:40 |
| jrosser | this would always dominate, perhaps thats why everything else seems never to make any difference | 15:41 |
| noonedeadpunk | might be. and eventually it kinda covers what mnaser has found with amount of variables | 15:41 |
| jrosser | i have tried putting ansibie local_tmp and remote_tmp in /dev/shm, all sorts of things to avoid disk IO | 15:42 |
| jrosser | but never anything makes it go faster | 15:42 |
| noonedeadpunk | yeah, because we had blocker even before connection | 15:43 |
| noonedeadpunk | *we have it eventually | 15:43 |
| *** cshen has quit IRC | 15:44 | |
| *** gyee has joined #openstack-ansible | 15:59 | |
| *** rpittau is now known as rpittau|afk | 16:00 | |
| noonedeadpunk | #startmeeting openstack_ansible_meeting | 16:00 |
| openstack | Meeting started Tue Jun 23 16:00:32 2020 UTC and is due to finish in 60 minutes. The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot. | 16:00 |
| openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 16:00 |
| *** openstack changes topic to " (Meeting topic: openstack_ansible_meeting)" | 16:00 | |
| openstack | The meeting name has been set to 'openstack_ansible_meeting' | 16:00 |
| noonedeadpunk | #topic office hours | 16:00 |
| *** openstack changes topic to "office hours (Meeting topic: openstack_ansible_meeting)" | 16:00 | |
| jrosser | o/ hello | 16:00 |
| noonedeadpunk | \o/ | 16:00 |
| noonedeadpunk | we had pretty huge progress last week thanks to jrosser, mgariepy and spatel :) | 16:01 |
| noonedeadpunk | so focal is ready (but don't want to merge) and centos8 is passing metal source scenario | 16:01 |
| jrosser | we were really on a roll last week | 16:02 |
| noonedeadpunk | and it's really pretty close | 16:02 |
| noonedeadpunk | yeah | 16:02 |
| noonedeadpunk | but we have train failing because of suse:( | 16:02 |
| noonedeadpunk | not sure if it's going to pass all in all, orwe should set it to non-voting? | 16:02 |
| jrosser | tbh i think we have so much else to do that it should go non-voting | 16:03 |
| noonedeadpunk | we eventually have lot's of stuff that ready for merging but fails on gates | 16:03 |
| jrosser | i've noticed a lot of failures on buster | 16:03 |
| noonedeadpunk | yeah, in lxc | 16:04 |
| spatel | noonedeadpunk: thank you for help me out with process, always happy to help you folks :) | 16:04 |
| jrosser | spatel: it is really great to see patches from you! | 16:04 |
| noonedeadpunk | but it feels it's something we should have fixed | 16:04 |
| noonedeadpunk | +2 | 16:04 |
| noonedeadpunk | also we have copied adjutant role from elastx https://opendev.org/openstack/openstack-ansible-os_adjutant | 16:05 |
| noonedeadpunk | but had really no time for it's integration | 16:05 |
| noonedeadpunk | and patching with all stuf we have | 16:05 |
| jrosser | so i think we should try to make a decision on this strategy plugin thing | 16:07 |
| jrosser | it still fails tests when just tested as a single line change | 16:07 |
| noonedeadpunk | I think we totally need to get it somehow working | 16:07 |
| noonedeadpunk | I mean that maybe we have broken tests at the moment | 16:07 |
| jrosser | i still don't really understand what the test is doing which fails | 16:07 |
| jrosser | dropping the paramiko fallback is also definately giving another 20% faster | 16:08 |
| noonedeadpunk | it didn't differ for me, but now wwe know why) | 16:08 |
| noonedeadpunk | bottleneck was before connection... | 16:09 |
| noonedeadpunk | actually we probably need to make paramiko configurable? as not sure as everyone use ssh keys | 16:09 |
| noonedeadpunk | 737543 looks really pretty fair to me | 16:10 |
| noonedeadpunk | but eventually I don't really like iteration over all invetory hosts... | 16:12 |
| jrosser | does it do that? | 16:13 |
| noonedeadpunk | no, it's what we have now and can get rid of with https://review.opendev.org/#/c/737220/9 | 16:13 |
| noonedeadpunk | I really can't find why we need pha | 16:14 |
| noonedeadpunk | and what use it... | 16:14 |
| jrosser | i was thinking it was like a reference | 16:14 |
| jrosser | so we add a thing to the object in the loop | 16:15 |
| jrosser | and pha is a reference to the object | 16:15 |
| noonedeadpunk | you mean LINEAR.display.verbose ? | 16:16 |
| noonedeadpunk | I'm not sure it's smth functional at all.... | 16:16 |
| jrosser | sorry have to go | 16:17 |
| jrosser | teatime | 16:17 |
| noonedeadpunk | yeah, sure | 16:17 |
| noonedeadpunk | looking at https://github.com/ansible/ansible/blob/devel/lib/ansible/plugins/strategy/__init__.py#L313 I have even less understanding what and why we're doing these loops... | 16:23 |
| noonedeadpunk | lol | 16:23 |
| noonedeadpunk | I guess I know what's wrong | 16:23 |
| noonedeadpunk | https://opendev.org/openstack/openstack-ansible-plugins/src/branch/master/strategy/linear.py#L212 this is probably supposed to be pha | 16:25 |
| noonedeadpunk | than it makes sense | 16:25 |
| *** dave-mccowan has joined #openstack-ansible | 16:27 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-plugins master: Optimise linear strategy access to physical host variables https://review.opendev.org/737543 | 16:28 |
| noonedeadpunk | and problem which was "resolved" with the patch introduced that looks pretty common :p https://opendev.org/openstack/openstack-ansible-plugins/commit/94c93e533f533206611ed70934b5819d7a318289 | 16:34 |
| *** markvoelker has joined #openstack-ansible | 16:40 | |
| *** markvoelker has quit IRC | 16:45 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-plugins master: Optimise linear strategy access to physical host variables https://review.opendev.org/737543 | 16:45 |
| noonedeadpunk | my judgement was wrong... so I still nee no impact from this cycle except logging things. In the meanwhile old code (which it replaced) was really altering behaviour | 16:50 |
| noonedeadpunk | #endmeeting | 16:51 |
| *** openstack changes topic to "Launchpad: https://launchpad.net/openstack-ansible || Weekly Meetings: https://wiki.openstack.org/wiki/Meetings/openstack-ansible || Review Dashboard: https://bit.ly/2SAcGAn" | 16:51 | |
| openstack | Meeting ended Tue Jun 23 16:51:33 2020 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 16:51 |
| openstack | Minutes: http://eavesdrop.openstack.org/meetings/openstack_ansible_meeting/2020/openstack_ansible_meeting.2020-06-23-16.00.html | 16:51 |
| openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/openstack_ansible_meeting/2020/openstack_ansible_meeting.2020-06-23-16.00.txt | 16:51 |
| openstack | Log: http://eavesdrop.openstack.org/meetings/openstack_ansible_meeting/2020/openstack_ansible_meeting.2020-06-23-16.00.log.html | 16:51 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-plugins master: DNM https://review.opendev.org/737583 | 16:56 |
| *** udesale_ has quit IRC | 17:05 | |
| *** spatel has quit IRC | 17:08 | |
| *** spatel has joined #openstack-ansible | 17:09 | |
| *** sshnaidm|ruck is now known as sshnaidm|afk | 17:24 | |
| *** mgariepy has quit IRC | 17:28 | |
| *** miloa has quit IRC | 17:31 | |
| *** mmethot has quit IRC | 17:36 | |
| *** cshen has joined #openstack-ansible | 17:41 | |
| *** mgariepy has joined #openstack-ansible | 17:41 | |
| *** cshen has quit IRC | 17:45 | |
| jrosser | noonedeadpunk: i think i agree that the current code that makes an assignment to 'pha' has no side effect at all | 17:51 |
| noonedeadpunk | but it's change makes tests fail:( | 17:52 |
| noonedeadpunk | argh | 17:52 |
| noonedeadpunk | just dnm passed CI https://review.opendev.org/#/c/737583/ | 17:52 |
| jrosser | hrrm | 17:52 |
| noonedeadpunk | I just don't see smth | 17:53 |
| jrosser | oohhhhhhhh | 17:53 |
| jrosser | this line https://github.com/openstack/openstack-ansible-plugins/blob/master/strategy/linear.py#L128 | 17:54 |
| jrosser | creates a new task var that is an empty dict | 17:54 |
| noonedeadpunk | oh | 17:54 |
| noonedeadpunk | and pha is the same object | 17:54 |
| jrosser | and i add a print here https://github.com/openstack/openstack-ansible-plugins/blob/master/strategy/linear.py#L210 | 17:54 |
| noonedeadpunk | yeah | 17:54 |
| *** markvoelker has joined #openstack-ansible | 17:54 | |
| jrosser | which shows it then populated with {'infra1': '10.31.128.12'} in task_vars['physical_host_addrs'] | 17:55 |
| noonedeadpunk | yeah, it's like python object symlinking lol | 17:55 |
| jrosser | so question is then, what on earth is the test case that fails trying to check | 17:56 |
| noonedeadpunk | so, if you're printing - what's the difference between self._variable_manager.get_vars(host=ph) and ph.get_vars() ? | 17:57 |
| jrosser | first one prints sooooooo much stuff it scrolls till i CTRL-C it | 17:58 |
| jrosser | like all the vars for everything, host/group blah blah i think | 17:58 |
| mgariepy | lol | 17:58 |
| jrosser | second one is printing this http://paste.openstack.org/show/795114/ | 17:59 |
| noonedeadpunk | just seems we miss smth in the second case... but not sure what and if we really need it... | 17:59 |
| *** markvoelker has quit IRC | 17:59 | |
| jrosser | i guess the only was is to run the functional test locally and see whats happening | 17:59 |
| jrosser | but yes i really don't know what it's doing | 17:59 |
| * jrosser end of day now | 18:00 | |
| *** cshen has joined #openstack-ansible | 18:01 | |
| *** cshen has quit IRC | 18:06 | |
| *** mmethot has joined #openstack-ansible | 18:10 | |
| *** cshen has joined #openstack-ansible | 18:10 | |
| *** markvoelker has joined #openstack-ansible | 18:11 | |
| *** markvoelker has quit IRC | 18:15 | |
| *** markvoelker has joined #openstack-ansible | 18:21 | |
| *** spatel has quit IRC | 18:29 | |
| *** markvoelker has quit IRC | 18:31 | |
| *** itandops has joined #openstack-ansible | 18:42 | |
| *** jawad_axd has joined #openstack-ansible | 18:50 | |
| *** cshen has quit IRC | 18:57 | |
| *** ebbex has quit IRC | 19:00 | |
| *** ebbex has joined #openstack-ansible | 19:02 | |
| *** cshen has joined #openstack-ansible | 19:03 | |
| *** cshen has quit IRC | 19:08 | |
| *** spatel has joined #openstack-ansible | 19:10 | |
| spatel | jamesdenton: Hey! i have question related OVN if you around | 19:11 |
| *** ebbex has quit IRC | 19:16 | |
| *** ebbex has joined #openstack-ansible | 19:17 | |
| *** itandops has quit IRC | 19:29 | |
| *** itandops has joined #openstack-ansible | 19:29 | |
| jamesdenton | spatel I'll try | 19:36 |
| spatel | jamesdenton: sorry i was away, now i am back. | 19:48 |
| spatel | Just hearing lots of buzz around OVN so trying to understand what advantage i will get over what we have currently? | 19:48 |
| jamesdenton | depends on what you're doing | 19:50 |
| jamesdenton | the advantage, as i understand it, is deprecation of python agents in favor of built-in functionality (built-in to OVS itself). There is still a lack of parity in some cases | 19:50 |
| jamesdenton | ML2/OVS will eventually be deprecated | 19:50 |
| spatel | oh! | 19:59 |
| spatel | You are saying ML2/LinuxBridge also eventually get deprecate? | 20:00 |
| jamesdenton | yes? i think that will fall out of main tree at some point, but remain a config option (separate repo) | 20:01 |
| spatel | Reason i am asking to understand where we are going from here? is OVN going to be a thing soon? | 20:01 |
| jamesdenton | OVN has been around for a few years now. Not sure where it's used in production, but i imagine it is. It's deployable with OSA, too, if you want to live on the edge. | 20:02 |
| spatel | so community trying to make OVN de facto deployment of openstack in coming future. | 20:02 |
| jamesdenton | OVN may become the default devstack driver, soon, too. | 20:02 |
| jamesdenton | yes, basically | 20:02 |
| spatel | hmm! i am just trying to convince myself to deploy OVN on my production. | 20:04 |
| spatel | if its very complicated then may wait until everyone start deploying | 20:04 |
| jamesdenton | i would wait | 20:04 |
| jamesdenton | are you using neutron routers? | 20:05 |
| spatel | +1 | 20:05 |
| spatel | No i am not using any software router. I am using all physical network | 20:05 |
| spatel | Because of performance issue. | 20:05 |
| spatel | last week we hot 140Gbps / 30million packet per second on our core router, i don't think neutron router can handle that rate | 20:06 |
| spatel | s/hot/hit/ | 20:06 |
| *** markvoelker has joined #openstack-ansible | 20:25 | |
| *** markvoelker has quit IRC | 20:29 | |
| spatel | jamesdenton: i am getting this error and trying to play with policy.json - {"message": "(rule:update_port and rule:update_port:binding:profile) is disallowed by policy" | 20:34 |
| jamesdenton | if you're not using L3 agent then i wouldn't push to insert OVN any time soon | 20:34 |
| spatel | update_port policy is already "update_port": "rule:admin_or_owner or rule:context_is_advsvc" | 20:34 |
| spatel | +1 | 20:35 |
| jamesdenton | what are you trying to do? | 20:35 |
| spatel | terraform to build sriov instances | 20:35 |
| *** also_stingrayza has joined #openstack-ansible | 20:35 | |
| spatel | look like terraform trying to touch some api of neutron which is not allowing to do | 20:36 |
| spatel | I am seeing this error on "queens" only | 20:36 |
| jamesdenton | hmm | 20:36 |
| spatel | I have stain which is working great | 20:37 |
| jamesdenton | that is weird. same policy in both places? | 20:37 |
| *** stingrayza has quit IRC | 20:37 | |
| spatel | yes that is odd | 20:37 |
| spatel | currently i have -> "update_port:binding:profile": "rule:admin_only" | 20:38 |
| spatel | thinking to change it "rule:admin_or_owner" | 20:38 |
| spatel | we don't need to restart anything right after updating policy.json right? | 20:40 |
| jamesdenton | can't recall | 20:40 |
| spatel | let me give it a try.. | 20:40 |
| spatel | jamesdenton: look like it works, not getting any error | 20:48 |
| spatel | let me try again and verify | 20:48 |
| jamesdenton | nice | 20:50 |
| spatel | yes not getting any error :) | 20:51 |
| spatel | policy.json doesn't need any restart. | 20:51 |
| spatel | jamesdenton: now question is does OSA support policy.json changes? | 20:54 |
| spatel | I did hand edit for now. | 20:55 |
| jamesdenton | might still work | 20:55 |
| jamesdenton | https://docs.openstack.org/project-deploy-guide/openstack-ansible/draft/app-advanced-config-override.html | 20:55 |
| spatel | oh yeah | 21:02 |
| spatel | thanks | 21:02 |
| *** spatel has quit IRC | 21:29 | |
| *** itandops has quit IRC | 21:37 | |
| *** jbadiapa has quit IRC | 21:38 | |
| *** markvoelker has joined #openstack-ansible | 22:25 | |
| *** markvoelker has quit IRC | 22:30 | |
| *** tosky has quit IRC | 22:42 | |
| *** cshen has joined #openstack-ansible | 22:54 | |
| *** cshen has quit IRC | 22:58 | |
| *** markvoelker has joined #openstack-ansible | 23:29 | |
| *** markvoelker has quit IRC | 23:34 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!