| opendevreview | Merged openstack/openstack-ansible stable/xena: Bump OpenStack-Ansible for Xena RC2 and release https://review.opendev.org/c/openstack/openstack-ansible/+/822259 | 00:45 |
|---|---|---|
| noonedeadpunk | at least now I know whom to ask question about ironic when they arise :p | 02:04 |
| noonedeadpunk | would be great to get vote for EOL of Ocata at least https://review.opendev.org/c/openstack/openstack-ansible/+/795690 | 02:09 |
| jrosser_ | would be nice to get ARA fixed - i pushed this after we talked about the runtime venv constraints the other day https://review.opendev.org/c/openstack/openstack-ansible/+/822088/3 | 07:44 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-os_keystone master: DNM - test db setup with logging enabled https://review.opendev.org/c/openstack/openstack-ansible-os_keystone/+/822391 | 07:44 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: [WIP] Update ansible-core to 2.12.1 https://review.opendev.org/c/openstack/openstack-ansible/+/822063 | 07:44 |
| jrosser_ | noonedeadpunk: you are right it is missing selinux https://paste.openstack.org/show/811790/ | 08:21 |
| noonedeadpunk | tbh I'm surpised it went that long | 08:37 |
| noonedeadpunk | have you reproduced that in aio? | 08:38 |
| jrosser_ | no, not done an AIO for this | 08:41 |
| noonedeadpunk | ok, will spawn then :) | 08:41 |
| noonedeadpunk | wonder if that selinux module would jsut work.... | 08:41 |
| jrosser_ | I think our current patches already use rsync for selinux in the runtime venv | 08:42 |
| noonedeadpunk | well, there're no selinux for 3.8? | 08:42 |
| noonedeadpunk | *python selinux bindings | 08:42 |
| noonedeadpunk | so we need either to build one or dunno | 08:43 |
| noonedeadpunk | (eventually centos explicitly said in docs they won't provide binding for 3.8) | 08:44 |
| jrosser_ | noonedeadpunk: as the db setup is delegated to oslodb_setup_host and uses oslodb_ansible_python_interpreter this looks more about the target python / selinux rather than the ansible controller? | 09:27 |
| noonedeadpunk | it's failing also for metal? | 09:28 |
| noonedeadpunk | well, yeah, might be. eventually we need to run services also in 3.8 python... | 09:28 |
| noonedeadpunk | 3.6 EOLing in 2 days | 09:28 |
| jrosser_ | super odd policy from RH again on this | 09:29 |
| jrosser_ | it was just the same with python3 on centos-7 | 09:30 |
| noonedeadpunk | yeah, and it's same | 09:43 |
| *** dviroel_ is now known as dviroel | 11:19 | |
| noonedeadpunk | jrosser_: do you think we should have https://review.opendev.org/c/openstack/ansible-role-systemd_service/+/816531 or just abandon as it brings a lot of complexity for arguable profit? | 12:04 |
| noonedeadpunk | it happened to us to be handy to use templated service outside of osa | 12:05 |
| opendevreview | James Denton proposed openstack/openstack-ansible-os_ironic master: Update Ironic Documentation https://review.opendev.org/c/openstack/openstack-ansible-os_ironic/+/822203 | 14:30 |
| noonedeadpunk | does anybody has any idea how that works? https://opendev.org/openstack/ansible-hardening/src/branch/master/tasks/main.yml#L55 | 14:40 |
| noonedeadpunk | shouldn't it work only with inlcude as there's a variable? | 14:41 |
| noonedeadpunk | ah, only variables from invetory can't be used, ok | 14:45 |
| jamesdenton | good to know | 14:55 |
| noonedeadpunk | #startmeeting openstack_ansible_meeting | 15:00 |
| opendevmeet | Meeting started Tue Dec 21 15:00:23 2021 UTC and is due to finish in 60 minutes. The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:00 |
| opendevmeet | The meeting name has been set to 'openstack_ansible_meeting' | 15:00 |
| noonedeadpunk | #topic office hours | 15:00 |
| noonedeadpunk | o/ | 15:00 |
| damiandabrowski[m] | hey! | 15:02 |
| spatel | jamesdenton morning | 15:04 |
| jamesdenton | hi | 15:04 |
| spatel | https://satishdotpatel.github.io/trex-load-generator/ | 15:04 |
| spatel | blog that out | 15:04 |
| jamesdenton | very nice, thank you. | 15:04 |
| jamesdenton | if you want to develop a test method i am happy to reproduce, and we can compare | 15:04 |
| jamesdenton | hi noonedeadpunk | 15:05 |
| spatel | Blog this for documentation, so i used VMware virtual machine. now i am going to do same on physical machine. | 15:05 |
| spatel | i am going to add TestPMD and L2fwd also so just for document. | 15:06 |
| opendevreview | Jeremy Stanley proposed openstack/openstack-ansible master: Prepare for stable/ocata EOL https://review.opendev.org/c/openstack/openstack-ansible/+/822500 | 15:08 |
| *** dviroel is now known as dviroel|lunch | 15:09 | |
| noonedeadpunk | So. | 15:10 |
| noonedeadpunk | 24.0.0 is about to be merged https://review.opendev.org/c/openstack/releases/+/822370/1 | 15:10 |
| noonedeadpunk | Hopefully it will be done today, but not 100% sure | 15:11 |
| noonedeadpunk | also stable/ocata branch has been dropped for roles, which results in failure of linters check https://review.opendev.org/c/openstack/openstack-ansible/+/795690 | 15:11 |
| jamesdenton | :thumbsup: | 15:12 |
| spatel | jamesdenton ironic question | 15:13 |
| opendevreview | Jeremy Stanley proposed openstack/openstack-ansible stable/ocata: Prepare for stable/ocata EOL https://review.opendev.org/c/openstack/openstack-ansible/+/822501 | 15:13 |
| spatel | in my dc i already have PXE vlan for kickstart. can i use same vlan for ironic pxe? | 15:14 |
| noonedeadpunk | so let's merge this one ^ | 15:14 |
| fungi | assuming it passes of course | 15:14 |
| noonedeadpunk | and get done with ocata finally | 15:15 |
| spatel | I am thinking i should shutdown older pxe but that is not going to happen soon i am thinking show i can run both together | 15:15 |
| jamesdenton | spatel let's hold off on that convo | 15:15 |
| spatel | copy that | 15:15 |
| fungi | per discussion in #opendev, odds are the new failures on stable/ocata are due to that branch already reaching eol and getting deleted from other repositories | 15:15 |
| noonedeadpunk | fungi: I wonder if we need noop for gates? | 15:15 |
| fungi | i left in one template for a job which seems to be passing | 15:15 |
| fungi | hopefully publish-openstack-sphinx-docs includes what i think it does there | 15:16 |
| fungi | i'm checking now | 15:16 |
| fungi | yeah, it's running build-openstack-sphinx-docs in check so should do the same in gate | 15:17 |
| noonedeadpunk | yeah, at least it does in master, indeed) | 15:17 |
| fungi | that's the job 795690 was still able to succeed | 15:17 |
| noonedeadpunk | thanks for helping! | 15:18 |
| fungi | well, by "it's running" i mean i can see it in progress in the status page | 15:18 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible stable/ocata: Switch Ocata to EOL https://review.opendev.org/c/openstack/openstack-ansible/+/795690 | 15:18 |
| *** akahat|ruck is now known as akahat|dinner | 15:18 | |
| noonedeadpunk | ok, then we can probably talk about centos stuff :) | 15:19 |
| fungi | and the status page indicates that ^ since rebase is now running just that one job as well | 15:19 |
| noonedeadpunk | as for Y we need to have py3.8 for ansible and for venvs | 15:20 |
| noonedeadpunk | which leads us to missing libselinux python bindings at very least | 15:20 |
| noonedeadpunk | jrosser_: epel seems better now for centos9 stream btw! https://dl.fedoraproject.org/pub/epel/9/Everything/x86_64/Packages/ | 15:21 |
| noonedeadpunk | including systemd-netwrokd :) | 15:21 |
| noonedeadpunk | and as you might all know py3.6 EOLing in a day or so. I guess at the end we will build libselinux for python from scratch... I'd hate doing that though, but no sure there's much choice here | 15:23 |
| fungi | are the prebuilt packages for 3.6 going to disappear? | 15:24 |
| noonedeadpunk | no, I don't think so | 15:24 |
| fungi | or you mean so that you can use newer python3-libselinux versions in the future | 15:24 |
| noonedeadpunk | I even not sure if they would dissapear fro pypi | 15:24 |
| fungi | the old versions wouldn't unless their authors explicitly hid/removed those | 15:25 |
| noonedeadpunk | So if we want to run ansible-core 2.12, which requires py3.8, then we have to get somehow python3.8-libselinux | 15:25 |
| noonedeadpunk | which is not provided by centos 8-stream | 15:26 |
| fungi | also if the maintainers of that project declare requires_python>=3.7 then pip should just pick the newest version which still declared support for 3.6 | 15:26 |
| noonedeadpunk | and redhat explicitly said they won't | 15:26 |
| opendevreview | Merged openstack/openstack-ansible stable/ocata: Prepare for stable/ocata EOL https://review.opendev.org/c/openstack/openstack-ansible/+/822501 | 15:26 |
| fungi | out of curiosity are you testing 3.8 because the pti says you should? because it actually says 3.6 and 3.9, doesn't mention 3.8 | 15:27 |
| noonedeadpunk | well, mostly because ubuntu 20.04 shipped with 3.8 | 15:28 |
| fungi | yep, that makes sense then | 15:28 |
| noonedeadpunk | and we don't build pyenv right now as then we would need to build pretty much of pre-packaged stuff | 15:28 |
| noonedeadpunk | so we're trying to leverage what distros ship | 15:28 |
| fungi | (though you can also use 3.9 on ubuntu 20.04 lts, it's just not the default python3) | 15:29 |
| noonedeadpunk | (for debian it's 3.9) | 15:29 |
| noonedeadpunk | oh ,well... | 15:29 |
| fungi | currently focal includes python3.9 (3.9.5-3ubuntu0~20.04.1) if you want to install it | 15:30 |
| noonedeadpunk | I wonder if they have python3.9 lxc binding hehe | 15:30 |
| noonedeadpunk | but yeah you're right | 15:30 |
| noonedeadpunk | we might indeed want to run py39 for focal | 15:30 |
| fungi | python3-lxc (1:3.0.4-1ubuntu6) requires python3 >= 3.8~ << 3.9, i don't see a separate build if it for 3.9 on focal | 15:31 |
| noonedeadpunk | I actually never looked how much projects are already running tests against 3.9 | 15:31 |
| fungi | so yeah, that could be a blocker | 15:31 |
| noonedeadpunk | but worth looking anyway, maybe there's smth in UCA for instance... | 15:33 |
| noonedeadpunk | but feels like indeed we will just run default py version there | 15:35 |
| noonedeadpunk | and with 22.04 it will be 3.9 or 3.10 maybe? | 15:35 |
| fungi | 3.10 as far as i know | 15:37 |
| fungi | there's a 3.10.1 backport due to land in focal-updates any day now according to motu | 15:37 |
| noonedeadpunk | oh, that's nice actually | 15:38 |
| noonedeadpunk | I still can't get used to that short life period of python versions after 2.7 :D | 15:38 |
| fungi | they didn't want to backport 3.10.0 to focal and were waiting for 3.10.1 to become available, which released a couple of weeks ago | 15:38 |
| opendevreview | Merged openstack/openstack-ansible stable/ocata: Switch Ocata to EOL https://review.opendev.org/c/openstack/openstack-ansible/+/795690 | 15:38 |
| noonedeadpunk | nice! ^ | 15:39 |
| fungi | congrats! out with the old, in with the new | 15:39 |
| noonedeadpunk | yeah, we haven't eol-ed stuff for a while | 15:43 |
| *** akahat|dinner is now known as akahat|ruck | 15:45 | |
| noonedeadpunk | #endmeeting | 15:51 |
| noonedeadpunk | #endmeeting | 15:51 |
| opendevmeet | Meeting ended Tue Dec 21 15:51:39 2021 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:51 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/openstack_ansible_meeting/2021/openstack_ansible_meeting.2021-12-21-15.00.html | 15:51 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/openstack_ansible_meeting/2021/openstack_ansible_meeting.2021-12-21-15.00.txt | 15:51 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/openstack_ansible_meeting/2021/openstack_ansible_meeting.2021-12-21-15.00.log.html | 15:51 |
| jamesdenton | spatel i would recommend not overlapping the two PXE environments. | 15:54 |
| spatel | ok | 15:54 |
| spatel | i have to spun up new vlan then | 15:55 |
| jamesdenton | on the ironic side, best i can tell, is that without any sort of ML2 driver or external automation, you're looking at needing a 'flat' setup where cleaning, provisioning, inspection, and post-deployment (i.e. tenant) use will occur on the same vlan/network | 15:55 |
| spatel | hmm | 15:56 |
| spatel | i am planning to play on VMware so i can do quick POC how does vlan and networking will look | 15:57 |
| jamesdenton | sure | 15:57 |
| spatel | jamesdenton quick question related my neutron-server memory consumption issue, look at this output and tel me what is wrong? - https://paste.opendev.org/show/811805/ | 15:57 |
| spatel | why i have duplicates proc, does it looks ok ? | 15:57 |
| jamesdenton | i use the 'neutron' setup w/ networking-generic-switch plugin, so the cleaning, provisioning, inspection networks are all different neutron provider networks, and the switchport config is managed automatically. | 15:57 |
| spatel | jamesdenton could you please blog that how about how do you running ironic :) | 15:58 |
| jamesdenton | yeah, i'll get there :) | 15:58 |
| spatel | +1 | 15:58 |
| jamesdenton | multiple api workers and rpc workers. looks normal | 15:59 |
| spatel | hmm ok, i thought i have something weird going on | 16:00 |
| jamesdenton | in later releases you will see that as 'neutron-server: api worker' | 16:00 |
| spatel | my neutron chewing up all 128GB memory in 2 week :( | 16:00 |
| jamesdenton | that's what cron is for | 16:00 |
| jamesdenton | :D | 16:00 |
| spatel | That is what i am planning to setup.. to kick it every week | 16:01 |
| jamesdenton | whatever it takes | 16:01 |
| spatel | i dont' have bandwidth to debug :( | 16:01 |
| jamesdenton | well you are also 5 releases behind | 16:01 |
| jamesdenton | so who knows | 16:01 |
| jamesdenton | (maybe not worth the efforts) | 16:02 |
| spatel | yes totally, i have Queens and wallaby running with same number of compute but they never ever hit more than 4gb | 16:02 |
| *** dviroel|lunch is now known as dviroel | 16:07 | |
| jrosser_ | noonedeadpunk: epel/9 missing lsyncd :( | 16:10 |
| noonedeadpunk | last lsyncd release was 4 years ago.... | 16:33 |
| noonedeadpunk | but github seems kind of alive | 16:33 |
| noonedeadpunk | most commits are kind of weeks ago | 16:33 |
| noonedeadpunk | I bet we had build from source code one day... | 17:02 |
| noonedeadpunk | https://review.opendev.org/c/openstack/openstack-ansible-repo_server/+/747911/1/tasks/repo_install_lsyncd.yml | 17:06 |
| noonedeadpunk | so we had actually same for centos 8:) | 17:07 |
| noonedeadpunk | I wonder if it makes sense to reach epel folks and ask if they plan to provide lsyncd :D | 17:07 |
| noonedeadpunk | we implemented with https://review.opendev.org/c/openstack/openstack-ansible-repo_server/+/709532 | 17:08 |
| noonedeadpunk | should be relatively easy to return that code back | 17:09 |
| noonedeadpunk | jrosser_: um, have you made a dnm patch for centos 9? As I can't find it for some reason... | 17:19 |
| noonedeadpunk | or you just played locally? | 17:20 |
| *** sshnaidm is now known as sshnaidm|afk | 17:34 | |
| noonedeadpunk | there are interesting stuff about epel I didn't know... | 17:45 |
| noonedeadpunk | So they build everything from scratch and maintainers should explicitly say they want their package to be in epel 9 for exaple before they will publish it | 17:45 |
| noonedeadpunk | So the process involve pinging maintainers to publish their packages with each new release | 17:46 |
| noonedeadpunk | https://docs.fedoraproject.org/en-US/epel/epel-package-request/ | 17:46 |
| noonedeadpunk | and that's why we saw just iptables at start :D | 17:46 |
| noonedeadpunk | I actually created request for lsyncd https://bugzilla.redhat.com/show_bug.cgi?id=2034693 | 17:46 |
| noonedeadpunk | and even regardles lsyncd is packaged for fedora 36, it might be absent for centos 9 epel even though centos 9 forked from fedora 34 (which also have lsyncd https://src.fedoraproject.org/rpms/lsyncd) | 17:48 |
| jrosser_ | noonedeadpunk: i just played locally but it failed on boostrap-aio due to systemd-networkd so there were really no patches to make | 18:28 |
| jrosser_ | gave up very quickly | 18:28 |
| noonedeadpunk | yeah, I do recall that... | 18:33 |
| jrosser_ | we're also missing the copr repo for lxc | 18:37 |
| jrosser_ | but i expect we could get a metal deployment done with building lsync from source | 18:37 |
| noonedeadpunk | well, for epel 8 it's not that bad lxc https://src.fedoraproject.org/rpms/lxc | 18:39 |
| noonedeadpunk | let me try to create another bug report :) | 18:40 |
| noonedeadpunk | let's see where this would lead us https://bugzilla.redhat.com/show_bug.cgi?id=2034709 | 18:44 |
| *** gmann is now known as gmann_afk | 19:51 | |
| *** dviroel is now known as dviroel|afk | 19:53 | |
| spatel | i like the idea of using lsyncd for source, we should use same method of both ubuntu and centos/el | 20:26 |
| spatel | these guys keep changing stuff on each release | 20:26 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!