| *** cshen has joined #openstack-ansible | 00:44 | |
| *** cshen has quit IRC | 00:49 | |
| *** watersj has quit IRC | 01:05 | |
| *** jawad_axd has joined #openstack-ansible | 01:12 | |
| *** threestrands has joined #openstack-ansible | 01:19 | |
| *** jawad_axd has quit IRC | 01:45 | |
| *** rh-jelabarre has quit IRC | 01:58 | |
| *** cshen has joined #openstack-ansible | 02:02 | |
| *** cshen has quit IRC | 02:07 | |
| *** cshen has joined #openstack-ansible | 04:03 | |
| *** cshen has quit IRC | 04:07 | |
| *** dpaclt has joined #openstack-ansible | 04:11 | |
| *** jcath has joined #openstack-ansible | 04:19 | |
| *** evrardjp has quit IRC | 04:33 | |
| *** evrardjp has joined #openstack-ansible | 04:33 | |
| *** jawad_axd has joined #openstack-ansible | 04:48 | |
| *** udesale has joined #openstack-ansible | 04:56 | |
| *** cheng1 has quit IRC | 06:00 | |
| *** cheng1 has joined #openstack-ansible | 06:02 | |
| *** djhankb has quit IRC | 06:08 | |
| *** jeh has quit IRC | 06:10 | |
| *** vesper has quit IRC | 06:18 | |
| *** LuckyClover has quit IRC | 06:18 | |
| *** vesper11 has joined #openstack-ansible | 06:18 | |
| *** cheng1 has quit IRC | 06:25 | |
| *** cheng1 has joined #openstack-ansible | 06:28 | |
| openstackgerrit | Carlos Augusto da Silva Martins proposed openstack/openstack-ansible-os_horizon stable/stein: Add drop_down_choices to Horizon https://review.opendev.org/734753 | 06:36 |
|---|---|---|
| *** jeh has joined #openstack-ansible | 06:43 | |
| *** cheng1 has quit IRC | 07:02 | |
| *** cheng1 has joined #openstack-ansible | 07:04 | |
| *** cshen has joined #openstack-ansible | 07:07 | |
| openstackgerrit | Jonathan Rosser proposed openstack/ansible-role-systemd_service master: Revert "Build out the PrivateNetwork function for services" https://review.opendev.org/734598 | 07:19 |
| *** rpittau|afk is now known as rpittau | 07:21 | |
| *** tosky has joined #openstack-ansible | 07:29 | |
| *** itandops has joined #openstack-ansible | 07:34 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Remove git repo haproxy backend https://review.opendev.org/734587 | 08:13 |
| janno | I'd appreciate if someone can comment on my task before commiting it: https://pastebin.com/E1S0UicZ | 08:14 |
| janno | especially with regard to disabled Cinder services | 08:14 |
| janno | Is it necessary to take them into account? | 08:15 |
| janno | Are they a thing in production environments? | 08:15 |
| *** kaiokmo has joined #openstack-ansible | 08:18 | |
| * noonedeadpunk missing context | 08:18 | |
| noonedeadpunk | eventually you may push a patch to gerrit, it's like a nicer place to leave reviews:) | 08:19 |
| noonedeadpunk | you can modify it easily afterwards | 08:19 |
| janno | alright | 08:19 |
| noonedeadpunk | also I can say I have disabled cinder service... | 08:20 |
| noonedeadpunk | But rly not sure I got what problem we're trying to eliminate here.... | 08:21 |
| janno | I had the problem during the upgrade from 19.0.1 to 19.1.1 that one cinder-service was down. The upgrade therefore failed during db online_data_migrations with some unspecific error | 08:24 |
| noonedeadpunk | ok, and how does this task would resolve it? | 08:25 |
| janno | and afair jrosser suggested that a check if all cinder services are available before doing the migrations makes sense | 08:25 |
| noonedeadpunk | So the thing is that it can be pre-upgrade check | 08:26 |
| janno | it just does a check if all services are up. if not it fails | 08:26 |
| noonedeadpunk | So all depends on what you're going to patch | 08:26 |
| janno | tasks/cinder_db_sync.yml | 08:26 |
| jrosser | yeah we have a fair few people who get in trouble with cinder migrations, and it's really unobvious that you need all the cinder-volume services to be up | 08:26 |
| jrosser | the error message is totally unhelpful and you end up with a traceback in the log i think, and little clue about why it's breaking | 08:27 |
| janno | I'd add the task as first task to tasks/cinder_db_sync.yml | 08:27 |
| janno | a check before upgrading would not have helped in my case because the failed service was caused by the upgrade | 08:28 |
| noonedeadpunk | I think then we need assert or smth to display a valid error here | 08:29 |
| jrosser | this would also need to be robust to whatever happens during a fresh install too | 08:29 |
| noonedeadpunk | and eventually, we need online_data_migrations only during major upgrades | 08:29 |
| jrosser | janno: looking back through my notes when we encoutered similar things at upgrade, disabling the broken backend services was not enough to make the db migrarion work | 08:30 |
| jrosser | they had to either be deleted or fixed before you could proceed | 08:31 |
| noonedeadpunk | + | 08:31 |
| jrosser | also, good morning all :) | 08:32 |
| noonedeadpunk | \o/ | 08:32 |
| janno | in this case this check would be sufficient. but yes, you are right, I don't yet know how it behaves during fresh install | 08:32 |
| jrosser | janno: easiest thing so to make a patch and push it to gerrit, that will all get tested | 08:34 |
| noonedeadpunk | I'd add at least conditions the same as for online_data_migrations task and run_once | 08:36 |
| noonedeadpunk | or smth | 08:36 |
| noonedeadpunk | (just to run this task just once) | 08:37 |
| jrosser | i tried to put a bit more details in https://etherpad.opendev.org/p/openstack-ansible-ussuri | 08:40 |
| janno | this is already the case, because import_tasks: cinder_db_sync.yml is bound to the conditions | 08:43 |
| janno | I'll push it in a second | 08:43 |
| noonedeadpunk | yeah, ok, youre right | 08:43 |
| noonedeadpunk | anyway I guess https://opendev.org/openstack/openstack-ansible-os_cinder/src/branch/master/tasks/cinder_db_sync.yml#L26-L28 needed? | 08:43 |
| janno | might make sense | 08:45 |
| noonedeadpunk | Also would be cool to do this with 2 tasks - first your one, which would check and register, and second like this https://opendev.org/openstack/openstack-ansible-os_cinder/src/branch/master/tasks/main.yml#L28-L31 | 08:47 |
| noonedeadpunk | which would show valid error message with some kind of explanations why we fail and what to do | 08:47 |
| jrosser | noonedeadpunk: i was giving myself a bit of a break from the focal work, looks like upgrades are broken | 08:49 |
| jrosser | did you think that was to do with apt-pinning in galera? | 08:49 |
| noonedeadpunk | T-U upgrades? | 08:50 |
| noonedeadpunk | ok, we didn't merge https://review.opendev.org/#/c/733301/... | 08:53 |
| noonedeadpunk | Really need to look why functional tests randomly fail | 08:53 |
| noonedeadpunk | (in terms what to check to really wait for cluster sync completition status) | 08:53 |
| jrosser | upgrades on master, yes, like here https://review.opendev.org/#/c/732123/ | 08:54 |
| jrosser | oh right yes i see.... wonder whats wrong with all those ssl jobs..... | 08:55 |
| noonedeadpunk | "var/lib/mysql: found previous version 10.3\r\nFailed to stop mysql.service: Unit mysql.service not loaded.\r\n'systemctl stop mysql' returned 5\r\nThere is a MySQL server running, but we failed in our attempts to stop it." | 08:56 |
| noonedeadpunk | So it's not pinning but really just the process of mariadb upgrade fails... | 08:56 |
| noonedeadpunk | So we kind of check for wsrep_incoming_addresses in https://review.opendev.org/#/c/733301/1/tests/test-galera-server-functional.yml and sometimes it's ok, and sometimes it's not enough... | 08:57 |
| janno | I am already re-writing the task | 08:57 |
| *** poopcat has quit IRC | 09:03 | |
| janno | I'll test it again first before pushing it | 09:06 |
| *** poopcat has joined #openstack-ansible | 09:09 | |
| *** jbadiapa has joined #openstack-ansible | 09:10 | |
| *** spatel has joined #openstack-ansible | 09:13 | |
| *** spatel has quit IRC | 09:18 | |
| janno | next try, not tested yet: https://pastebin.com/masLb1fq | 09:20 |
| noonedeadpunk | janno: oh.... | 09:24 |
| noonedeadpunk | maybe we can just use block and rescue? | 09:24 |
| jcath | hello, jrosser. I try to test aio deployment of OSA today, when I go to apply setup_openstack.yml now, it reports "misspelled module name" when proceed to service_setup.yml of keystone. I found that module name in yml is openstack.cloud.os_keystone_domain, something like that, but at sub-dir under /opt/ansible-runtime, i see this path: /opt/ansible-runtime/lib/python3.6/site-packages/ansible/modules/cloud/openstack, it seems the name's order is | 09:25 |
| jcath | reversed? | 09:25 |
| noonedeadpunk | So in case online_data_migrations fails we go to rescue, check for service list and display reasonable mesaage? | 09:25 |
| noonedeadpunk | So we do like try\except | 09:25 |
| jrosser | jcath: you are using master? | 09:25 |
| noonedeadpunk | If we don't catch error with online migrations - we get rid of another task | 09:26 |
| jcath | yes, I think. I just git clone the openstack-ansible, and following the quickref of aio | 09:26 |
| jcath | bootstrap-ansible, bootstrap-aio .... | 09:26 |
| jrosser | jcath: i would recommend that you use either the stable/train, or a specific tag from the train branch | 09:26 |
| janno | noonedeadpunk: I'm not sure if broken cinder services can be the only reason for the task to fail. | 09:27 |
| jcath | oh, i c. | 09:27 |
| janno | noonedeadpunk: if there is another reason it might be misleading | 09:27 |
| jrosser | jcath: master is where we are working on the Victoria release, Ussuri branch is currently a release candidate and not finished, so Train is currently the best choice | 09:28 |
| jcath | try now... thanks, jrosser | 09:28 |
| noonedeadpunk | janno: so we can do smth like that http://paste.openstack.org/show/794555/ | 09:28 |
| noonedeadpunk | yes, this might be not the only reason. | 09:28 |
| jcath | how about stein, my last deployment with packstack is stein, is there much difference between stein/train? | 09:29 |
| noonedeadpunk | so we can probably add another fail task there with `when: failed_cinder_services.rc == 1` | 09:29 |
| noonedeadpunk | and some general messae that migrations are failed while all services are up | 09:30 |
| janno | http://paste.openstack.org/show/794556/ | 09:35 |
| noonedeadpunk | I think it's perfect now - as we really cover our issue with untrivial failure | 09:36 |
| noonedeadpunk | Also we can probably try to start down services as step 2 afterwards or figure ou some another workaround | 09:37 |
| janno | Before testing I need to re-install my test cluster. Should I wait with pushing until I tested it locally or shall I push now? | 09:37 |
| noonedeadpunk | I'd say push it | 09:37 |
| noonedeadpunk | and will see in CI what happens :) | 09:37 |
| noonedeadpunk | I just missed when section, hope it's still there :p) | 09:38 |
| *** amalrajgenocidex has joined #openstack-ansible | 09:55 | |
| amalrajgenocidex | I was deploying magnum and encountered below bug. It exists in all stable/train and ussuri. Is there any fix for it? | 09:56 |
| amalrajgenocidex | https://storyboard.openstack.org/#!/story/2007591#comment-159235 | 09:56 |
| amalrajgenocidex | Trying to label master node with node-role.kubernetes.io/master="" | 09:57 |
| amalrajgenocidex | k8s-calico-coreos-3dljpm3zmgow-master-0 --patch '{"metadata": {"labels": {"node-role.kubernetes.io/master": ""}}}' | 09:57 |
| openstackgerrit | Jan Marquardt proposed openstack/openstack-ansible-os_cinder master: Added checks for failed cinder services https://review.opendev.org/734778 | 09:59 |
| janno | damn | 10:00 |
| janno | become and become_user in "Perform online data migrations" is redundant | 10:01 |
| noonedeadpunk | just edit patch, do git add .; git commit --amend; git review | 10:03 |
| noonedeadpunk | (if you need an edit ofc) | 10:03 |
| openstackgerrit | Jan Marquardt proposed openstack/openstack-ansible-os_cinder master: Added checks for failed cinder services https://review.opendev.org/734778 | 10:03 |
| *** cshen has quit IRC | 10:03 | |
| noonedeadpunk | and delete trailing whitespaces :p | 10:04 |
| openstackgerrit | Jan Marquardt proposed openstack/openstack-ansible-os_cinder master: Added checks for failed cinder services https://review.opendev.org/734778 | 10:05 |
| janno | better? :P | 10:06 |
| noonedeadpunk | yep:) | 10:08 |
| *** sshnaidm|afk is now known as sshnaidm | 10:14 | |
| jrosser | amalrajgenocidex: that is a bug in magnum rather than openstack-ansible i think? | 10:18 |
| *** rpittau is now known as rpittau|bbl | 10:19 | |
| jrosser | amalrajgenocidex: it seems they are working on a patch for that from the storyboard link | 10:20 |
| jrosser | once that merges into magnum and is backported to the branch you want OSA can deploy that for you | 10:20 |
| *** cshen has joined #openstack-ansible | 10:55 | |
| amalrajgenocidex | Thanks jrosser . I will wait for the bug fix then :( (y) | 11:05 |
| *** cshen has quit IRC | 11:07 | |
| *** watersj has joined #openstack-ansible | 11:12 | |
| *** threestrands has quit IRC | 11:16 | |
| *** zigo has quit IRC | 11:41 | |
| *** amalrajgenocidex has quit IRC | 11:41 | |
| openstackgerrit | amolkahat proposed openstack/openstack-ansible-os_tempest master: Added tempest ironic resources setup. https://review.opendev.org/720705 | 11:44 |
| janno | there have been two failed Zuul checks for my commit, but they look unrelated | 11:45 |
| *** zigo has joined #openstack-ansible | 11:55 | |
| *** rh-jelabarre has joined #openstack-ansible | 11:57 | |
| *** rpittau|bbl is now known as rpittau | 12:05 | |
| *** cshen has joined #openstack-ansible | 12:18 | |
| openstackgerrit | Sagi Shnaidman proposed openstack/openstack-ansible-os_tempest master: Use ansible openstack collection https://review.opendev.org/718639 | 12:29 |
| *** itandops_ has joined #openstack-ansible | 12:30 | |
| mgariepy | morning everyone | 12:32 |
| *** itandops has quit IRC | 12:34 | |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible master: Remove /usr/bin/python on bionic for CI jobs https://review.opendev.org/731476 | 12:51 |
| janno | when doing fresh install, I, now and then, stumble across missing openstacksdk messages: http://paste.openstack.org/show/794564/ | 12:52 |
| janno | I know how to fix this, but I am wondering if this is an unknown issue or is it fixed in newer versions? I am currently installing 19.0.1, but I saw this during upgrade from Stein to Train as well | 12:54 |
| *** udesale_ has joined #openstack-ansible | 12:54 | |
| mgariepy | how are the gate check prioritized to start ? are they starting depending on the time they take to complete or its somewhat random when the job is scheduled ? | 12:55 |
| *** udesale has quit IRC | 12:57 | |
| *** d34dh0r53 has quit IRC | 13:01 | |
| *** d34dh0r53 has joined #openstack-ansible | 13:07 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_horizon master: Compile mod-wsgi module for CentOS 7 https://review.opendev.org/734608 | 13:07 |
| noonedeadpunk | jrosser: have you looked closer on upgrade jobs stuff? | 13:07 |
| noonedeadpunk | as was going to look at it | 13:07 |
| jrosser | no, i've not looked at that today | 13:09 |
| mgariepy | jrosser, the depends on for the auto stuff seems to fix the bionic metal check on https://review.opendev.org/731476 | 13:12 |
| jrosser | janno: the tasks you paste there are delegated to a service setup host? | 13:12 |
| jrosser | oh err you have 10.10.10.3 and 10.10.10.1 there, do those addresses make sense? | 13:12 |
| jrosser | mgariepy: do you think we need both? i.e 'auto' will still prefer /usr/bin/python(2) if it's present? | 13:12 |
| * jrosser slightly losing the plot with this :) | 13:13 | |
| mgariepy | auto will prefer python3 is present for bionic | 13:14 |
| mgariepy | i think | 13:14 |
| janno | jrosser: yes, they are delegated to a service setup host | 13:14 |
| jrosser | janno: have you also set the python interpreter for the service setup host? | 13:15 |
| mgariepy | the auto will take the config in the base.yml file https://github.com/ansible/ansible/blob/stable-2.9/lib/ansible/config/base.yml#L1435-L1450 | 13:15 |
| *** d34dh0r53 has quit IRC | 13:15 | |
| jrosser | janno: i have a patch not yet merged to make this the default btw https://review.opendev.org/#/c/733854/ | 13:15 |
| mgariepy | let me look at the code once more | 13:16 |
| janno | jrosser: no, I don't set openstack_service_setup_host_python_interpreter | 13:17 |
| jrosser | mgariepy: i think what i mean is that your patch is passing for metal/bionic on it's own https://review.opendev.org/#/c/734672/ | 13:17 |
| jrosser | so mine is probably not needed | 13:17 |
| jrosser | janno: i think that is quite possibly the issue, openstacksdk will be installed into the utility venv in the utility host, not into the host python | 13:18 |
| openstackgerrit | Carlos Augusto da Silva Martins proposed openstack/openstack-ansible-os_horizon stable/stein: Horizon configuration to support domains dropdown https://review.opendev.org/734834 | 13:20 |
| mgariepy | https://github.com/ansible/ansible/blob/stable-2.9/lib/ansible/executor/interpreter_discovery.py#L192 | 13:23 |
| mgariepy | so, if the exac version for the distro is not there ie: centos7, it will take the config from centos6, dito for ubuntu versions. | 13:23 |
| mgariepy | so indeed we should not need to remove python from the CI. | 13:24 |
| *** cheng1 has quit IRC | 13:31 | |
| *** cshen has quit IRC | 13:35 | |
| *** cheng1 has joined #openstack-ansible | 13:36 | |
| *** cshen has joined #openstack-ansible | 13:44 | |
| *** cshen has quit IRC | 13:45 | |
| *** jawad_axd has quit IRC | 13:47 | |
| *** cheng1 has quit IRC | 13:50 | |
| *** cheng1 has joined #openstack-ansible | 13:52 | |
| *** udesale_ has quit IRC | 14:10 | |
| *** sshnaidm is now known as sshnaidm|bbl | 14:23 | |
| *** cheng1 has quit IRC | 14:38 | |
| *** cheng1 has joined #openstack-ansible | 14:41 | |
| *** also_stingrayza has joined #openstack-ansible | 14:44 | |
| *** stingrayza has quit IRC | 14:48 | |
| *** cheng1 has quit IRC | 15:29 | |
| *** cheng1 has joined #openstack-ansible | 15:33 | |
| johnsom | Hi OSA folks! We are getting some reports of folks installing Train with OSA getting a non-Train version of octavia-dashboard. Looking at the SHA in https://github.com/openstack/openstack-ansible/blob/stable/train/playbooks/defaults/repo_packages/openstack_services.yml#L309 I can't find that SHA in octavia-dashboard (or anywhere really). This looks like some sort of automated script set this. | 15:43 |
| johnsom | Can someone point me to where this script lives? | 15:43 |
| johnsom | I'm trying to figure out where that SHA came from. | 15:43 |
| johnsom | My guess is that this SHA is bogus and people are ending up with the master branch version. | 15:44 |
| noonedeadpunk | oh.... | 15:44 |
| noonedeadpunk | But this sha is eventually the latest commit for stable/train? https://opendev.org/openstack/octavia-dashboard/commits/branch/stable/train | 15:45 |
| noonedeadpunk | So https://opendev.org/openstack/octavia-dashboard/commit/74044687e0ef5b74853258a0f9ce2be7320e88f0 is valid link | 15:46 |
| johnsom | Oh, now that is strange, I re-checked it out and found it. | 15:47 |
| johnsom | Nevermind then, I will still look into this. It's odd as the code for the error they are getting is only in Ussuri and newer. | 15:48 |
| noonedeadpunk | but.... | 15:49 |
| johnsom | Hmm, that is so odd. The search on github still doesn't find that sha. | 15:49 |
| noonedeadpunk | I think I know what the issue is | 15:49 |
| johnsom | Ah, I am all ears | 15:49 |
| noonedeadpunk | let me just check the idea... | 15:50 |
| noonedeadpunk | yeah.... | 15:50 |
| johnsom | BTW, this has been reported here: https://bugs.launchpad.net/openstack-ansible/+bug/1881874 | 15:51 |
| openstack | Launchpad bug 1881874 in octavia-dashboard (Ubuntu) "Unable to launch Loadbalancer from Horizon Dashboard " [Undecided,Incomplete] | 15:51 |
| noonedeadpunk | Was just looking at it today :) | 15:52 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_horizon master: Add octavia version to constraints https://review.opendev.org/734881 | 15:53 |
| noonedeadpunk | johnsom: ^ | 15:53 |
| *** gyee has joined #openstack-ansible | 15:54 | |
| johnsom | noonedeadpunk Ah, interesting. Thanks! | 15:54 |
| noonedeadpunk | we just missed it:( | 15:54 |
| noonedeadpunk | dpaclt: I guess you was facing this issue | 15:55 |
| *** rpittau is now known as rpittau|afk | 16:07 | |
| openstackgerrit | amolkahat proposed openstack/openstack-ansible-os_tempest master: Added tempest ironic resources setup. https://review.opendev.org/720705 | 16:08 |
| *** cheng1 has quit IRC | 16:08 | |
| *** cheng1 has joined #openstack-ansible | 16:09 | |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: Convert host and container journal collection to python https://review.opendev.org/734883 | 16:11 |
| noonedeadpunk | found issue with upgrade jobs... | 16:15 |
| *** maharg101 has quit IRC | 16:19 | |
| *** maharg101 has joined #openstack-ansible | 16:26 | |
| jrosser | oh excellent | 16:28 |
| noonedeadpunk | in terms of glera, unfortunatelly | 16:29 |
| noonedeadpunk | (not how to make them use current patches and depends-on unfortunatelly) | 16:29 |
| * noonedeadpunk should fix jobs itelfs as well... | 16:30 | |
| openstackgerrit | James Denton proposed openstack/openstack-ansible-os_ironic master: WIP - Add support for uefi booting https://review.opendev.org/734886 | 16:31 |
| *** gyee has quit IRC | 16:42 | |
| dpaclt | noonedeadpunk yeah i was facing issue with stable/train and i just upgraded octavia alone to ussuri and the issue was fixed | 16:50 |
| dpaclt | And yeah i raised the bug and also mentioned it in storyboard too | 16:50 |
| dpaclt | Not exactly with stable/train but with branch 20.1.1 as i was unable to deploy with stable/train OSA at that time | 16:51 |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible master: auto-detect interpreter. https://review.opendev.org/734672 | 16:52 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-galera_server master: Fix Mariadb 10.3 -> 10.4 upgrade path https://review.opendev.org/734888 | 16:53 |
| *** jcath has quit IRC | 16:53 | |
| noonedeadpunk | dpaclt: eventually using tags even recommended way... | 16:54 |
| noonedeadpunk | and that patch should fix horizon octavia dashboard | 16:54 |
| noonedeadpunk | https://review.opendev.org/734881 | 16:54 |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible master: auto-detect interpreter. https://review.opendev.org/734672 | 16:54 |
| dpaclt | noonedeadpunk so this https://review.opendev.org/734881 should fix octavia dashboard in train ? | 16:55 |
| noonedeadpunk | I hope it should | 16:56 |
| noonedeadpunk | It would be the same for train | 16:56 |
| dpaclt | ohh ok Thanks | 16:59 |
| *** gyee has joined #openstack-ansible | 17:08 | |
| jrosser | need another vote on this https://review.opendev.org/#/c/732064/ to unblock stein | 17:11 |
| mgariepy | looking | 17:11 |
| jrosser | cool thanks | 17:12 |
| mgariepy | done :) haha | 17:12 |
| mgariepy | wasn't that hard to review | 17:12 |
| mgariepy | jrosser, how do we change the interpreter patch for debian only ? | 17:14 |
| mgariepy | can we do | omit for the interpreter var ? | 17:14 |
| jrosser | which one? | 17:14 |
| mgariepy | the one you did for 20.04 | 17:14 |
| mgariepy | would probably need it for debian | 17:15 |
| mgariepy | https://review.opendev.org/#/c/733948/ | 17:15 |
| jrosser | the decision about which interpreter to use is here https://review.opendev.org/#/c/733948/17/test-ansible-functional.sh | 17:16 |
| mgariepy | i like the auto one, but for debian it's not going to get a backport on 2.9 | 17:16 |
| mgariepy | https://review.opendev.org/#/c/733948/17/test-prepare-containers.yml can we do omit here ? | 17:17 |
| jrosser | so yesterday i mentioned a hybrid approach | 17:17 |
| jrosser | in my patch you could do auto | 17:17 |
| jrosser | except for debian where we set it specifically | 17:17 |
| jrosser | so just change the logic in test-ansible-functional.sh to be that | 17:18 |
| jrosser | omit only works on module parameters | 17:18 |
| mgariepy | ok | 17:19 |
| jrosser | sort of like this maybe http://paste.openstack.org/show/794594/ | 17:19 |
| mgariepy | ok i'll modify your patch to test that. | 17:20 |
| jrosser | hopefully we can use the auto keyword anywhere we like | 17:21 |
| mgariepy | i dont think it's possible | 17:21 |
| mgariepy | interpreter_python can be auto in either ENV or default section of ansible.cfg | 17:22 |
| jrosser | i was thinking we could either template ansible.cfg, or carry a python3 specific ansible.cfg | 17:22 |
| mgariepy | You can still set ansible_python_interpreter to a specific path at any variable level (for example, in host_vars, in vars files, in playbooks, etc.). Setting a specific path completely disables automatic interpreter discovery; Ansible always uses the path specified. | 17:22 |
| jrosser | i think there is a step somewhere that symlinks ansible.cfg to the right place | 17:22 |
| jrosser | mgariepy: maybe you can do something here to just write the value we want into ansible.cfg? https://github.com/openstack/openstack-ansible-tests/blob/master/test-ansible-env-prep.sh#L248-L258 | 17:24 |
| mgariepy | wow. the doc is really confixung | 17:24 |
| mgariepy | https://docs.ansible.com/ansible/latest/reference_appendices/config.html#envvar-ANSIBLE_PYTHON_INTERPRETER | 17:25 |
| mgariepy | i'll do some tests i guess. | 17:25 |
| *** jbadiapa has quit IRC | 17:30 | |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: Convert host and container journal collection to python https://review.opendev.org/734883 | 17:33 |
| *** dpaclt has quit IRC | 17:37 | |
| *** sshnaidm|bbl is now known as sshnaidm | 17:40 | |
| noonedeadpunk | jrosser: I'm a bit stuck with galera..... trying to find what's was first chicken or egg.... | 17:43 |
| mgariepy | probably the chicken | 17:44 |
| noonedeadpunk | So the thing is, that when we run https://opendev.org/openstack/openstack-ansible-galera_server/src/branch/master/tasks/galera_install_apt.yml#L81 it drops galera-server package | 17:44 |
| noonedeadpunk | which is needed here https://opendev.org/openstack/openstack-ansible-galera_server/src/branch/master/tasks/galera_server_upgrade_pre.yml#L22 | 17:45 |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible master: auto-detect interpreter. https://review.opendev.org/734672 | 17:49 |
| *** cheng1 has quit IRC | 17:50 | |
| *** cheng1 has joined #openstack-ansible | 17:50 | |
| jrosser | why does installing packages remove them? | 17:53 |
| noonedeadpunk | Dependencies? http://paste.openstack.org/show/794595/ | 17:53 |
| noonedeadpunk | we're installing 10.4 client over 10.3... | 17:54 |
| noonedeadpunk | It updates mariadb-common which depends on server.... | 17:54 |
| noonedeadpunk | *on which server depends | 17:54 |
| mgariepy | would probably need to do the client and server pkg at the same time ? | 17:55 |
| noonedeadpunk | So when we check for the service it already do not exist, so we don;'t stop it and fail during new package isntallation:( | 17:55 |
| noonedeadpunk | so we don't run galera_client part against galera hosts? | 17:56 |
| noonedeadpunk | hm, yeah, that's interesting | 17:56 |
| mgariepy | jrosser, do you mind if i merge your patch in mine ? | 18:00 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-galera_server master: Fix Mariadb 10.3 -> 10.4 upgrade path https://review.opendev.org/734888 | 18:00 |
| jrosser | mgariepy: sure go for it | 18:00 |
| noonedeadpunk | jrosser: https://review.opendev.org/#/c/734608/ worked for me in sandbox | 18:04 |
| noonedeadpunk | https://199.204.45.225/ | 18:04 |
| noonedeadpunk | mmmm | 18:05 |
| jrosser | indeed mmmmm | 18:05 |
| noonedeadpunk | It's more like network stuff:) | 18:06 |
| noonedeadpunk | or oom... | 18:06 |
| *** jbadiapa has joined #openstack-ansible | 18:06 | |
| jrosser | how does that work? | 18:08 |
| jrosser | is it a binary shipped as a pip module or does it build it? | 18:08 |
| noonedeadpunk | It builds it | 18:08 |
| noonedeadpunk | and then provides path to it | 18:08 |
| noonedeadpunk | that's why we need httpd-devel.... | 18:09 |
| noonedeadpunk | can go another patch and try to replace with uwsgi | 18:09 |
| noonedeadpunk | (it was just easier so strated from it) | 18:09 |
| jrosser | depends if we see a need to switch to uwsgi, like if there is benefit from changing things | 18:10 |
| noonedeadpunk | (restored vm access and put your keys for centos user on it) | 18:11 |
| noonedeadpunk | just in case | 18:11 |
| noonedeadpunk | regarding benefit... maybe it will be more about unification... but a bit scared about like proxying things to uwsgi... | 18:12 |
| noonedeadpunk | all headers and stuff... | 18:12 |
| noonedeadpunk | Ok, will try uwsgi option as well then and we'll have options to choose from | 18:13 |
| jrosser | it's always been a bit surprising the apache/nginx dual operation for keystone | 18:15 |
| noonedeadpunk | but don't really see anything wrong with building that bodule... It's build with apx, so it takes seconds to finish build | 18:15 |
| jrosser | if it's working, and we drop it anyway at some point with centos-8 coming on, it's not too bad | 18:15 |
| noonedeadpunk | for centos8 we're gonna build lsyncd :p | 18:16 |
| jrosser | swings/roundabouts :) | 18:16 |
| jrosser | i spent some time today messing with the log collection | 18:17 |
| jrosser | turns out its been too long since i did python | 18:17 |
| noonedeadpunk | oh, I miss that part also nowadays... Need some hands on osome py stuff as well... | 18:19 |
| jrosser | i just diregarded making it py2/3 | 18:19 |
| jrosser | need to address that as it's not going to work on centos7 | 18:19 |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible-tests master: auto-detect interpreter https://review.opendev.org/734676 | 18:20 |
| noonedeadpunk | ah, let's gonna drop it once get centos8... | 18:20 |
| noonedeadpunk | so whatever | 18:20 |
| noonedeadpunk | btw https://review.opendev.org/#/c/734888/ - upgrade worked on another sandbox now, but wonderign what about functional tests... | 18:21 |
| jrosser | oh yes the "when is the cluster ready" problem | 18:22 |
| jrosser | i was thinking maybe we should install container 1,2,3 with serial=1 | 18:23 |
| jrosser | perhaps it's just not good bringing the entire cluster up all at the same time | 18:23 |
| * noonedeadpunk goes to ask mariadb folks | 18:23 | |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible-tests master: auto-detect interpreter https://review.opendev.org/734676 | 18:24 |
| noonedeadpunk | but we do up start cluster with 1 container and then 2 others join it iirc | 18:24 |
| noonedeadpunk | so it's not actually all at the same time... | 18:24 |
| jrosser | mgariepy: test-ansible-env-prep.sh is called from test-ansible-functional.sh, do we need to set those vars in both places? | 18:25 |
| noonedeadpunk | we can actually wait until wsrep_cluster_size == 3 | 18:26 |
| mgariepy | ho.. | 18:26 |
| mgariepy | no i guess. | 18:26 |
| mgariepy | i was just trying to be clever haha | 18:26 |
| mgariepy | but failed miserably.. | 18:26 |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible-tests master: auto-detect interpreter https://review.opendev.org/734676 | 18:27 |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible-tests master: auto-detect interpreter https://review.opendev.org/734676 | 18:29 |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible-galera_server master: Use temporary repository for MariaDB 10.4.14 for Ubuntu focal https://review.opendev.org/733946 | 18:32 |
| mgariepy | hmm how do you get rid of the merge conflict ? | 18:33 |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible-tests master: auto-detect interpreter https://review.opendev.org/734676 | 18:34 |
| openstackgerrit | Marc Gariépy (mgariepy) proposed openstack/openstack-ansible-tests master: auto-detect interpreter https://review.opendev.org/734676 | 18:52 |
| *** sshnaidm is now known as sshnaidm|afk | 19:05 | |
| openstackgerrit | Merged openstack/openstack-ansible stable/stein: Rewrite CI provided apt sources to be http instead of https on bionic https://review.opendev.org/732064 | 19:43 |
| *** cheng1 has quit IRC | 19:48 | |
| *** cheng1 has joined #openstack-ansible | 19:49 | |
| *** cheng1 has quit IRC | 20:43 | |
| *** cheng1 has joined #openstack-ansible | 20:44 | |
| *** spatel has joined #openstack-ansible | 21:03 | |
| *** maharg101 has quit IRC | 21:08 | |
| *** maharg101 has joined #openstack-ansible | 21:11 | |
| *** spatel has quit IRC | 21:32 | |
| *** kaiokmo has quit IRC | 21:35 | |
| *** krah has quit IRC | 21:42 | |
| *** krah has joined #openstack-ansible | 21:45 | |
| *** threestrands has joined #openstack-ansible | 21:55 | |
| *** spatel has joined #openstack-ansible | 21:56 | |
| *** rh-jelabarre has quit IRC | 22:10 | |
| *** spatel has quit IRC | 22:34 | |
| *** itandops_ has quit IRC | 22:51 | |
| *** djhankb has joined #openstack-ansible | 23:02 | |
| *** tosky has quit IRC | 23:15 | |
| *** cheng1 has quit IRC | 23:57 | |
| *** cheng1 has joined #openstack-ansible | 23:59 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!