| *** gvincent has joined #ara | 08:24 | |
| *** jungleslow has quit IRC | 08:39 | |
| ssbarnea | dmsimard: https://review.openstack.org/#/c/627882/ is ready, contains all fixed needed for making all jobs pass | 08:42 |
|---|---|---|
| openstackgerrit | Sorin Sbarnea proposed openstack/ara master: Eliminate use of ansible include https://review.openstack.org/627875 | 09:26 |
| *** sshnaidm is now known as sshnaidm|afk | 11:48 | |
| *** sshnaidm|afk is now known as sshnaidm | 12:43 | |
| *** dougbtv has joined #ara | 13:20 | |
| *** jrist has joined #ara | 15:17 | |
| *** jrist has quit IRC | 15:22 | |
| *** jrist has joined #ara | 15:26 | |
| dmsimard | ssbarnea: lgtm, thanks for those | 15:51 |
| ssbarnea | dmsimard: consider it my cheap xmas present | 15:53 |
| dmsimard | I'd love to chat with tripleo folks about ara 1.0 at some point in the near future | 15:53 |
| dmsimard | I don't have any plans for supporting the backend/API on python 2 (just die already) but the callback can be made py2 without much effort -- especially the one that uses requests to send stuff over http | 15:54 |
| ssbarnea | sure. i suspect it involves some breaking changes or is just marketing? | 15:54 |
| dmsimard | 1.0 is a nothing short of a full rewrite | 15:55 |
| ssbarnea | oops... | 15:55 |
| *** spiette has quit IRC | 15:56 | |
| dmsimard | ssbarnea: I need to update the live demo (and in fact, I need to write a role to automatically redeploy it) but https://api.demo.recordsansible.org/api/v1/ | 15:56 |
| ssbarnea | if the api is still compatible it should not be big dead, AFAIK we only it call it few times so no reason to worry. but I am sure we will do it with py2 for a longer time. | 15:56 |
| dmsimard | no such thing as an API for 0.x, everything talks to the database directly | 15:57 |
| ssbarnea | for me 0,x api == cli usage, is still an api. | 15:57 |
| dmsimard | sure | 15:57 |
| dmsimard | ssbarnea: the new API allows you to do stuff like this though: https://ara-server.readthedocs.io/en/latest/using_api_clients.html#example-api-usage | 15:58 |
| dmsimard | which I'm pretty excited about | 15:59 |
| *** spiette has joined #ara | 16:00 | |
| openstackgerrit | Merged openstack/ara master: Fix testing warnings and makes future one treated as errors https://review.openstack.org/627882 | 16:07 |
| openstackgerrit | David Moreau Simard proposed openstack/ara-web master: WIP: Run a CI job that builds and publishes the web dashboard https://review.openstack.org/625666 | 16:12 |
| ssbarnea | dmsimard: if you can also make a release with the recent changes it would be great. last one was back in september. | 16:16 |
| dmsimard | ssbarnea: yes, we're due for a 0.16.2 | 16:17 |
| dmsimard | There was also a deprecation fix from the ara generate html command that landed | 16:17 |
| dmsimard | I was supposed to spend some time on ara during the holidays but ended up doing other (still productive!) things unrelated to ara :p | 16:18 |
| openstackgerrit | Merged openstack/ara master: Eliminate use of ansible include https://review.openstack.org/627875 | 16:19 |
| ssbarnea | dmsimard: do you happen to know what caused https://bugs.launchpad.net/tripleo/+bug/1792622 ? | 17:17 |
| openstack | Launchpad bug 1792622 in tripleo "Ara fails to install in some of tripleo jobs with ImportError: 'module' object has no attribute 'check_specifier'" [Medium,Fix released] - Assigned to Rafael Folco (rafaelfolco) | 17:17 |
| ssbarnea | nevermind, i found | 17:18 |
| dmsimard | ssbarnea: never seen that error before :p | 17:19 |
| ssbarnea | dmsimard: is caused by very old setuptools, in case you... | 17:20 |
| dmsimard | ssbarnea: the 1.0 implementation might look similar to the current middleware approach in 0.x -- except the web interface has an API client and expects an API endpoint and I have absolutely no interest in rewriting all the API logic in javascript :p | 17:55 |
| dmsimard | so what might end up happening is: 1) save database 2) write middleware for a "central" API backend to arbitrarily load those databases (not unlike what we have right now for /ara-report/) 3) have a static interface query that API with that database loaded (it doesn't matter where the interface is hosted, it's stateless) | 17:57 |
| ssbarnea | dmsimard: having one server would be a good enough solution for the time being. if we endup with someone with extensive js skills, they are welcome to redoit in pure js. | 17:57 |
| ssbarnea | yep, you just described what I was trying to say. | 17:57 |
| dmsimard | so the challenge is figuring out how to get the backend to load these arbitrary databases in a remotely safe and secure way | 17:58 |
| dmsimard | Right now it looks like this: https://github.com/openstack/ara/blob/master/ara/wsgi_sqlite.py | 17:59 |
| dmsimard | but it also relies on the database being located on the same filesystem as the application | 17:59 |
| dmsimard | If we're sending databases to swift, though, we'd need to download them first ? | 18:00 |
| ssbarnea | obviously, download them and cache them. is unlikely that they will change. | 18:00 |
| dmsimard | A clever hack would be to trick the application into thinking the database is actually on the same server with something like S3QL | 18:01 |
| dmsimard | Which "mounts" swift as a filesystem | 18:01 |
| ssbarnea | ara related but unrelated to this, https://review.openstack.org/#/c/606182/6 | 18:02 |
| ssbarnea | sshnaidm: ^ | 18:02 |
| *** sshnaidm is now known as sshnaidm|afk | 18:05 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!