| *** herald85 has joined #ara | 08:22 | |
| *** gnupyx has quit IRC | 12:23 | |
| dmsimard | apollo13: first shot at basic security things http://logs.openstack.org/81/632181/2/check/build-openstack-sphinx-docs/4cbb1bc/html/security.html | 13:52 |
|---|---|---|
| dmsimard | it doesn't contain the remote_user things | 13:53 |
| dmsimard | but if it's good enough, it's something we can iterate on | 13:53 |
| *** Pilou has joined #ara | 13:55 | |
| openstackgerrit | David Moreau Simard proposed openstack/ara-server master: First iteration of authentication and security docs https://review.openstack.org/632181 | 13:58 |
| dmsimard | Pilou: it's working for me right now with devel against master of ara-server, ara-clients and ara-plugins | 14:01 |
| apollo13 | dmsimard: still fighting with my VMs and xslt :D | 14:03 |
| dmsimard | Pilou: http://paste.openstack.org/show/743086/ | 14:03 |
| apollo13 | you don't think that one could just export an OVA from virtualbox and import in VMWare | 14:03 |
| apollo13 | I currently have a 50 lines xslt transforming the format so VMWare can actually read it | 14:03 |
| dmsimard | apollo13: qemu-img doesn't have something to convert from a format to the other ? | 14:03 |
| apollo13 | for ovf itself? doubt it | 14:04 |
| apollo13 | that is for disk images | 14:04 |
| dmsimard | interestingly docs from openstack https://docs.openstack.org/image-guide/convert-images.html | 14:04 |
| dmsimard | oh, OVA is what ? the VM manifest ? | 14:04 |
| apollo13 | yeah that is disks, not the metadata file | 14:04 |
| dmsimard | gotcha | 14:04 |
| apollo13 | ova is a tar of ovf + manifest + disks | 14:04 |
| apollo13 | and ovf is an xml describing the machine config | 14:05 |
| dmsimard | yeah like libvirt.xml :p | 14:05 |
| apollo13 | just worse :รพ | 14:05 |
| dmsimard | I still remember when openstack got rid of xml | 14:06 |
| dmsimard | everyone wanted to kill it with fire | 14:06 |
| dmsimard | because supporting both json and xml simultaneously was a pita | 14:07 |
| dmsimard | apollo13: do you have an opinion on how this mechanic might work in practice ? https://github.com/openstack/ara-clients/blob/master/ara/clients/utils.py#L22-L34 | 14:08 |
| dmsimard | That code is probably wrongly instanciating both clients as Pilou pointed out, my fault | 14:08 |
| dmsimard | It was more or less an experiment of how we might cleanly provide a client or another | 14:09 |
| dmsimard | Another question would be -- do we return an instance of the class or the class itself ? | 14:09 |
| Pilou | dmsimard: with this patch https://paste.fedoraproject.org/paste/8DDR1jxV5MZpKOFslgFQqw , i got this error: https://paste.fedoraproject.org/paste/rjV2wAw2YVjE0L2LPJ2oZA | 14:10 |
| dmsimard | Pilou: yeah that patch was a PoC -- api.demo.recordsansible.org isn't actually open for POSTs | 14:12 |
| dmsimard | Pilou: if you set the API_CLIENT to offline instead of http with api.demo, it might actually work | 14:13 |
| dmsimard | I had the integration test runner work locally | 14:14 |
| dmsimard | We need to improve a bit how the client handles errors like those | 14:14 |
| openstackgerrit | Merged openstack/ara-server master: Docs: SECRET_KEY is now randomly generated if it's not set https://review.openstack.org/632119 | 14:15 |
| dmsimard | oh nice | 14:16 |
| dmsimard | since zuul knows about github.com/ansible/ansible, we can use depends-on on PRs :D | 14:17 |
| apollo13 | dmsimard: yes it will instantiate both | 14:20 |
| apollo13 | though you could wrap it into a partial object and then call () afterwards again | 14:20 |
| dmsimard | meh, I don't have a strong opinion | 14:21 |
| dmsimard | I wrote it like that because that's how I do switch cases in python | 14:21 |
| apollo13 | https://dpaste.de/CZiZ | 14:21 |
| dmsimard | I don't know yet what the other client implementations will look like | 14:21 |
| apollo13 | this would fix it | 14:21 |
| apollo13 | if you could figure out how to handle the kwargs | 14:21 |
| dmsimard | I could just go back to the old if/else | 14:22 |
| dmsimard | eh | 14:22 |
| apollo13 | hehe, yeah | 14:22 |
| dmsimard | https://github.com/openstack/ara-clients/blob/b1dc3a0944d15d4c6db995d39b90e78abfd3578f/ara/clients/utils.py#L22-L28 | 14:22 |
| dmsimard | it doesn't need to be complicated | 14:22 |
| apollo13 | that said, if all clients supported **kwargs in __init__ you wouldn't have a problem | 14:22 |
| apollo13 | right, I was 1 second to late :) | 14:23 |
| dmsimard | I think I'd rather us return an instanciated client instead of the class -- it's sort of the entire point of the method | 14:23 |
| dmsimard | otherwise people can just import AraHttpClient or AraOfflineClient if they want | 14:24 |
| apollo13 | you didn't read what I wrote | 14:24 |
| apollo13 | you can do that, you just need to move () after the switch | 14:24 |
| apollo13 | like my link had | 14:24 |
| dmsimard | I read what you wrote, sorry | 14:24 |
| dmsimard | sort of replying out of context | 14:24 |
| apollo13 | that said the if is better than the switch | 14:24 |
| dmsimard | yeah I'll change it | 14:25 |
| apollo13 | the switch hides errors from the constructor | 14:25 |
| openstackgerrit | David Moreau Simard proposed openstack/ara-clients master: Use an if/else for returning the client in get_client https://review.openstack.org/632474 | 14:27 |
| dmsimard | Pilou: ^ | 14:28 |
| dmsimard | Pilou: was there a particular reason you mentioned we might want to return the class instead of an instance ? | 14:28 |
| dmsimard | Pilou: like I mentioned a bit earlier above, the point of get_client is sort of to return the instance -- otherwise people that know what they're doing can import AraHttpClient or AraOfflineClient directly | 14:29 |
| Pilou | dmsimard: nope, this way seems good :) | 14:29 |
| dmsimard | ok, cool | 14:29 |
| dmsimard | Pilou: btw 1.0 isn't tested against devel yet -- I caught that issue with 0.x integration tests | 14:29 |
| dmsimard | I'm going to add devel coverage soon. | 14:30 |
| Pilou | unrelated, is ARA compatible with Python 3.6 only ? | 14:30 |
| dmsimard | Pilou: f-strings are pretty nice and they landed in >=3.6 | 14:31 |
| dmsimard | I'm not sure if we use other >=3.6 exclusive features | 14:31 |
| Pilou | For the Ansible integration tests, there are not many docker containers which provides Python > 3.5 | 14:32 |
| Pilou | (ubuntu1804, fedora28 and fedora29) | 14:35 |
| dmsimard | really ? all of those most definitely ship with >3.5 | 14:35 |
| dmsimard | unless you mean only those are >3.5 | 14:35 |
| dmsimard | f29 ships with at least 3.7, f28 3.6 | 14:36 |
| dmsimard | 1804 is probably 3.6 | 14:36 |
| dmsimard | Pilou: thanks for spending some time to do these tests btw :) | 14:38 |
| dmsimard | apollo13: easy +2 on https://review.openstack.org/#/c/632474/ :p | 14:38 |
| Pilou | yes, those are >3.5 | 14:38 |
| dmsimard | ok, thanks | 14:39 |
| dmsimard | FWIW, support for <3.6 will likely come in the form of the callback and client | 14:39 |
| dmsimard | I can't commit to fully supporting py2 but in theory there could be a py2 compatible callback and client implementation | 14:39 |
| apollo13 | ?? | 14:40 |
| apollo13 | I somewhat doubt that | 14:40 |
| dmsimard | apollo13: what do you mean ? | 14:40 |
| apollo13 | not for the offline client though | 14:40 |
| dmsimard | apollo13: yes, it would be just http | 14:40 |
| dmsimard | it would probably not be too much work to make the http client work with py2 | 14:41 |
| apollo13 | it is currently impossible | 14:41 |
| apollo13 | since we are using namespaces which are py3 only | 14:41 |
| dmsimard | good point | 14:42 |
| dmsimard | we can figure something out | 14:42 |
| dmsimard | I am very much wondering if we should merge everything back into openstack/ara at some point | 14:42 |
| dmsimard | there's pros and cons to each approach | 14:43 |
| dmsimard | managing issues/bugs/triage/patches across more than one repo is a con :p | 14:44 |
| openstackgerrit | Merged openstack/ara-clients master: Use an if/else for returning the client in get_client https://review.openstack.org/632474 | 14:44 |
| dmsimard | I guess I am left mostly wondering how long 0.x will go for | 14:45 |
| dmsimard | It's already been mostly in maintenance mode for >1 year without a lot of new development | 14:46 |
| openstackgerrit | David Moreau Simard proposed openstack/ara-infra master: WIP: New ARA 1.0 integration jobs https://review.openstack.org/630303 | 15:45 |
| dmsimard | This is cool: https://github.com/NotAProfessionalDeveloper/ansible-tools/tree/master/ara_host_history | 15:49 |
| dmsimard | but it's also one of the reasons why the new API is so important | 15:49 |
| *** openstackgerrit has quit IRC | 15:51 | |
| *** openstackgerrit has joined #ara | 16:17 | |
| openstackgerrit | David Moreau Simard proposed openstack/ara-infra master: WIP: New ARA 1.0 integration jobs https://review.openstack.org/630303 | 16:17 |
| dmsimard | Maybe we should have a version of https://docs.ansible.com/ansible/latest/community/other_tools_and_programs.html to point out community projects around ara | 16:34 |
| dmsimard | like the one I linked above | 16:34 |
| openstackgerrit | David Moreau Simard proposed openstack/ara-infra master: WIP: New ARA 1.0 integration jobs https://review.openstack.org/630303 | 16:36 |
| *** herald85 has quit IRC | 17:15 | |
| *** ssbarnea|rover has joined #ara | 18:03 | |
| *** ssbarnea|bkp2 has quit IRC | 18:04 | |
| *** mgariepy has quit IRC | 21:06 | |
| openstackgerrit | David Moreau Simard proposed openstack/ara-infra master: WIP: New ARA 1.0 integration jobs https://review.openstack.org/630303 | 22:08 |
| openstackgerrit | David Moreau Simard proposed openstack/ara-infra master: WIP: New ARA 1.0 integration jobs https://review.openstack.org/630303 | 22:15 |
| openstackgerrit | David Moreau Simard proposed openstack/ara-plugins master: Fix typo when ara_record retrieves options from the callback https://review.openstack.org/632579 | 22:35 |
| openstackgerrit | David Moreau Simard proposed openstack/ara-infra master: WIP: New ARA 1.0 integration jobs https://review.openstack.org/630303 | 22:37 |
| ara-slack | mrz: anyone have ARA working with an RDS-backed postgres instance? | 22:43 |
| ara-slack | jneville: @jneville has joined the channel | 22:47 |
| *** mmercer has quit IRC | 22:50 | |
| dmsimard | @mrz I've seen users of postgre before but not RDS | 22:53 |
| dmsimard | I believe someone attempted to use the mysql variant (aurora?) but it ended up being less expensive to run it on a VM -- something about the amount of max connections if I remember correctly | 22:54 |
| dmsimard | Pilou: just got the first successful integration job of 1.0 against devel :) | 22:57 |
| dmsimard | http://logs.openstack.org/03/630303/17/check/ara-integration-fedora-devel/b01e308/ara-report/ | 22:58 |
| openstackgerrit | Merged openstack/ara-plugins master: Fix typo when ara_record retrieves options from the callback https://review.openstack.org/632579 | 23:05 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!