| Sage905 | I think it may have been a result of me using pre-built packages. I've switched back to source, and am trying again. | 00:02 |
|---|---|---|
| Sage905 | Just in case anyone else runs into it, I tracked it down to this: https://github.com/eventlet/eventlet/issues/781 | 01:13 |
| Mohaa | Hi | 07:52 |
| hamidlotfi_ | Hi mohaa | 07:52 |
| Mohaa | (((: | 07:52 |
| Mohaa | How are you Hamid? | 07:53 |
| hamidlotfi_ | Thank you, | 07:53 |
| Mohaa | I'm going to edit an interface ip of a router in the context of OVN. There's no namespace, so no working `ip netns exec qrouter-<ID> ip ...` | 07:53 |
| Mohaa | Should I deal with the southband DB? | 07:56 |
| *** dviroel_ is now known as dviroel | 11:47 | |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Update haproxy healthcheck options https://review.opendev.org/c/openstack/openstack-ansible/+/887285 | 12:41 |
| ncuxo | is it okay to run openstack as the hypervisor for your orga? https://www.youtube.com/watch?v=wsy9OY-ot7E&pp=ygUVb3BlbnN0YWNrIHdpc2ggSSBrbmV3 according to this video you shouldn't do that. My plan was to have a router and ntp server all the rest will come with openstack and machines provisioned on openstack, like FreeIPA as idm as an example | 12:44 |
| jrosser | ncuxo: you've asked us a very similar question before - it entirely depends on your use case | 12:53 |
| ncuxo | jrosser: sorry I don't remember the answer ... | 12:55 |
| jrosser | you should probably not run services inside your openstack which are fundamental to deploying it | 12:55 |
| jrosser | like dns, version control, the place you run ansible from, etc etc | 12:55 |
| ncuxo | and you told me the only service required for openstack was ntp so far I remember | 12:56 |
| jrosser | or whatever automation you use to provision hardware | 12:56 |
| jrosser | right, but the OSA documentation kind of assumes you've already got a solution to booting/managing/monitoring your hardware | 12:56 |
| ncuxo | ilo ? | 12:57 |
| jrosser | sure, use whatever you need to | 12:57 |
| ncuxo | and the deployment host from where you deploy the stack | 12:57 |
| jrosser | like i say it depends | 12:57 |
| ncuxo | however I haven't seen anything about monitoring. | 12:58 |
| jrosser | i have a large DMZ that sits behind our openstack with prometheus, grafana, ELK, PXEboot, DNS....... loads and loads of stuff | 12:58 |
| jrosser | and building that is outside the scope of openstack-ansible, hence why I say it entirely depends on your use case | 12:59 |
| anskiy | ncuxo: how much hardware are you planning to use for you openstack deployment? | 12:59 |
| jrosser | i think what you're really wanting to do is have no supporting infrastructure and get a whole "organisation" running by bringing up an openstack? | 13:00 |
| ncuxo | I wanted to start with 3 servers with all in deployment and then add specialised nodes as needed | 13:00 |
| ncuxo | jrosser: exactly | 13:00 |
| jrosser | right - so if you want to PXEboot, how would you do that? | 13:00 |
| ncuxo | ironic with virtual media on ilo I've read it is supported | 13:01 |
| jrosser | but ironic is a service that runs in openstack | 13:01 |
| jrosser | you have to break the catch-22 of having a pile of unconfigured hardware and bootstrap stuff from nothing | 13:01 |
| ncuxo | so as soon as I have the control plane I can add more nodes | 13:01 |
| jrosser | imho its two different things to build and maintain an openstack environment, and the build services on it | 13:02 |
| ncuxo | yeah ... then I really can't evade skipping support infra and hypervisors then :( | 13:03 |
| jrosser | if you want to use virtual media and ilo to provision the hardware thats completely fine | 13:03 |
| jrosser | so long as that meets whatever specific requirements you have, then it's good | 13:04 |
| ncuxo | jrosser: thanks for the clarifications | 13:05 |
| jrosser | sorry this is really difficult to answer | 13:05 |
| jrosser | because generally there are existing opinions on all this stuff in the organisations where openstack is deployed | 13:06 |
| jrosser | so being flexible and able to fit in with as many of those as possible is the approach OSA takes | 13:06 |
| jrosser | rather than have specific approaches baked in | 13:07 |
| jrosser | just be really careful deploying services inside your cloud that it comes to rely on iteslf, perhaps don't use FreeIPA for openstack authentication if you host it in the cloud itself | 13:08 |
| jrosser | similar with DNS, classic mistake | 13:09 |
| ncuxo | I was really looking at openstack for dns, dhcp and LB services otherwise I would have gone to baremetal k8s and install openstack on top of that | 13:17 |
| ncuxo | since I still will have to provision the DNS, DHCP, LB, PXE outside of k8s as well | 13:17 |
| jrosser | regardless of openstack vs. k8s try not to make circular things | 13:22 |
| jrosser | you will struggle to use openstack.example.com FQDN in your config before deployment if the DNS serving that zone is in your yet-to-be-deployed openstack/k8s | 13:23 |
| ncuxo | there is /etc/hosts :D poor mans dns :P | 13:24 |
| ncuxo | well regardless for either solution I need external services so yeah, no way around that | 13:25 |
| jrosser | yeah, to take the DNS example further, if you wanted to host zones which were not needed for your deployment inside a VM, thats a different matter and could be ok | 13:27 |
| jrosser | thats really what i mean about the "build/maintain/monitor platform" parts being a different problem to the "build services with" part | 13:27 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Update haproxy healthcheck options https://review.opendev.org/c/openstack/openstack-ansible/+/887285 | 14:16 |
| Sage905 | I switched from distro to source, and I no longer encounter the error I mentioned the other day. | 14:35 |
| Sage905 | I do, however, have another error that I haven't been able to figure out, yet: https://gist.github.com/ptoal/58bd6c3bf81e2ac60b6ddb22c9a5e420 | 14:38 |
| jrosser | Sage905: which release are you using? | 14:44 |
| jrosser | Sage905: unrelated but did you say the other day that you are using centos9? | 14:47 |
| Sage905 | @jrosser Git checkout aa558cc36 | 14:47 |
| Sage905 | Yes. Stream 9 | 14:47 |
| jrosser | so on centos9 all i can do is give you RH own words as warning "CentOS Stream may seem like a natural choice to replace CentOS Linux, but it is not designed for production use. It is intended as a development platform for Red Hat partners and others that want to participate and collaborate in the Red Hat Enterprise Linux ecosystem" | 14:47 |
| Sage905 | This is a lab environment. | 14:47 |
| jrosser | well regardless, they break it all the time is our experince of running CI jobs on it | 14:48 |
| jrosser | so `aa558cc36` is the tip of master branch, which means that you are running the code that we use as the development version for the *next* release | 14:49 |
| Sage905 | Well, for full disclosure, I am a Red Hatter. I don't want to run RDO / Openstack, because I was looking for something lighter. | 14:49 |
| jrosser | ok cool :) | 14:49 |
| Sage905 | Ah. Okay. I should have noticed that. | 14:50 |
| jrosser | our c9s jobs are currently all non-voting because, you guessed it, it's broken again | 14:50 |
| jrosser | but the error you pasted looks like an incompatibility between openstack SDK and the ansible collection for openstack | 14:51 |
| jrosser | stable/2023.1 would be a good branch to try for something recent but supposedly stable | 14:53 |
| ncuxo | Sage905: also there is no path of deploying rdo because tripleo is deprecated | 14:53 |
| jrosser | ^ openstack-ansible in distro mode does this | 14:53 |
| Sage905 | ncuxo: That was what I had thought, based on my research. Not looking forward to an openshift-based control plane in the short-term... Seems pretty heavy. | 14:55 |
| Sage905 | Okay, I'll revert my snapshot, switch to stable/2023.1, and try again. | 14:55 |
| jrosser | on stable/2023.1 we run a rocky linux / distro job and it passes CI, so that should be pretty close if you really need centos+rdo packages | 14:58 |
| jrosser | otherwise install from source is the default | 14:58 |
| Sage905 | Yeah, so this is just for my homelab, to build demos, and test code, etc. | 15:02 |
| damiandabrowski | if any has some spare time, we have 2 important patches waiting for reviews | 15:03 |
| Sage905 | I have VMW running on it right now, but I thought it would be fun (strange definition, perhaps) to replace it with OSP | 15:03 |
| damiandabrowski | 1. needed to unblock rocky jobs https://review.opendev.org/c/openstack/openstack-ansible-rabbitmq_server/+/887592 | 15:03 |
| damiandabrowski | 2. fix rally install https://review.opendev.org/c/openstack/openstack-ansible-os_rally/+/887528 | 15:03 |
| opendevreview | Merged openstack/openstack-ansible-rabbitmq_server master: Use wildcards to specify rabbit/erlang versions https://review.opendev.org/c/openstack/openstack-ansible-rabbitmq_server/+/887592 | 17:21 |
| mgariepy | hmm isn't 20 req / 10s a bit on the low side for haproxy ? | 17:33 |
| mgariepy | or maybe more for horizon ? | 17:35 |
| Sage905 | jrosser: Alas, with stable/2023.1 in source mode, I get the same error. | 17:37 |
| jrosser | mgariepy: if you think we should increase it then sure, i have noticed occasional issues with horizon and rate limit | 17:40 |
| mgariepy | i'll run it a bit like that maybe it's my shift-R to refresh that makes the issue :) | 17:43 |
| jrosser | really this is to keep a lid on bots hammering the external api endpoints | 17:44 |
| jrosser | so ideally horizon should have a more relaxed limit | 17:44 |
| jrosser | perhaps we can be smarter with it | 17:46 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-os_heat master: DNM - test heat deployment https://review.opendev.org/c/openstack/openstack-ansible-os_heat/+/887731 | 17:48 |
| jrosser | Sage905: ^ lets see what happens with that | 17:48 |
| mgariepy | of course.. https://review.opendev.org/c/openstack/nova/+/882914 | 18:03 |
| Sage905 | jrosser: Perhaps I'm missing something, but that looks like it's just an empty file named 'test' ? | 18:06 |
| mgariepy | how's 2023.1 doing ? | 18:06 |
| mgariepy | any big issue ? | 18:06 |
| Sage905 | Ah.. But there's this: https://bugs.launchpad.net/openstack-ansible/+bug/2024407 | 18:08 |
| mgariepy | huh. https://linuxcontainers.org/lxd/ | 18:09 |
| opendevreview | Merged openstack/openstack-ansible-os_rally master: Include proper commit in rally_upper_constraints_url https://review.opendev.org/c/openstack/openstack-ansible-os_rally/+/887528 | 19:15 |
| NeilHanlon | mgariepy: uh | 19:33 |
| mgariepy | indeed NeilHanlon | 19:34 |
| NeilHanlon | can we have one week, damnit? | 19:35 |
| mgariepy | yeah.. last year ;p | 19:36 |
| NeilHanlon | 😅 | 19:40 |
| mgariepy | well. | 19:44 |
| mgariepy | so.. hyperv ? | 19:44 |
| mgariepy | or goats.. | 19:44 |
| *** dviroel_ is now known as dviroel | 19:48 | |
| Sage905 | It's been a minute since I've used an actual IRC client... | 20:28 |
| ptoal | Thanks for the help. I managed to get it working. | 23:51 |
| ptoal | I applied this patch: https://review.opendev.org/c/openstack/openstack-ansible/+/887513 | 23:52 |
| ptoal | It took me a bit to figure out that I had to re-run scripts/bootstrap-ansible.sh to get it to take. | 23:52 |
| ptoal | Of course, I have a new problem, where I get an error that "something went wrong" when I try to log in, but at least it's a new problem! :) | 23:55 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!