| eandersson | This is the old shady code, but maybe a different path https://github.com/openstack/designate/tree/master/designate/network_api | 02:09 |
|---|---|---|
| johnsom | That is the old "sink" based neutron/nova integration right? Yeah, that should probably get deprecated if it hasn't been. | 02:11 |
| johnsom | lol, no warning signs there "GreenThreadPoolExecutor" | 02:12 |
| johnsom | Copied partially from nova | 02:12 |
| eandersson | https://github.com/openstack/designate/blob/stable/queens/designate/network_api/neutron.py - this was before I re-wrote all of that :p | 02:13 |
| eandersson | I don't think the neutron integration actually uses this. frickler probably knows | 02:14 |
| johnsom | I am not dis-ing the async, just green threads in general. grin | 02:14 |
| eandersson | hehe yea that was just a work around because of some unit tests issues | 02:15 |
| eandersson | because monkey patching wasn't done right | 02:15 |
| eandersson | would be nice to get ride of eventlet at some point | 02:16 |
| johnsom | Yes | 02:16 |
| eandersson | Pretty sure there are some issues with the wsgi implementation as well | 02:16 |
| eandersson | where it ends up using eventlet even thou it shouldn't be | 02:17 |
| eandersson | but maybe I fixed that already | 02:17 |
| johnsom | We went with cotyledon in Octavia | 02:17 |
| eandersson | I almost did when I re-wrote all the base frameworks for designate | 02:17 |
| eandersson | but some code I just didn't understand was messing with me | 02:18 |
| eandersson | and wanted to make sure to stay closer to other projects for easier maintainability | 02:18 |
| eandersson | since we didn't have a lot of contributors at the time | 02:18 |
| johnsom | Yeah, I can't yet sign up for major overhauls like that, but it's in the back of my mind. | 02:20 |
| frickler | I'm all for deprecating the PTR API, neutron integration works the other way round (neutron being the active part instead of designate) and those two are incompatible | 05:25 |
| opendevreview | Dr. Jens Harbott proposed openstack/designate master: Improve wording for validation error messages https://review.opendev.org/c/openstack/designate/+/820026 | 13:44 |
| opendevreview | Michael Johnson proposed openstack/designate master: Improve wording for validation error messages https://review.opendev.org/c/openstack/designate/+/820026 | 15:45 |
| johnsom | Quick fix for the unit tests | 15:45 |
| frickler | johnsom: thx, I was in a hurry to push this before I went into a meeting and didn't check locally | 16:37 |
| johnsom | Yeah, no problem. I thought you might have been done for the day. | 16:37 |
| frickler | makes me wonder though whether the tests should actually check for the exact wording. and if they should, are we then even allowed to change it? | 16:38 |
| frickler | maybe add a reno at least? | 16:39 |
| johnsom | Yeah, I tend to not test for the wording. It just leads to test churn as things evolve | 16:39 |
| johnsom | Though, hopefully we aren't changing strings that often | 16:40 |
| frickler | this is an interesting failure I think https://0c3226f8ab8cd996497a-d2c025200b43adc89ab93bbaf8ea8562.ssl.cf5.rackcdn.com/820026/2/check/designate-ipv6-only-pdns4/05f0ce7/index.html | 17:12 |
| frickler | a) I think the test should fail immediately when the status reaches DELETE/ERROR and not wait another 10 minutes for something to happen | 17:13 |
| frickler | b) looking at the pdns logs, it seems like the delete is triggered when pdns is still busy processing the initial AXFR | 17:14 |
| frickler | not sure what we can do about that, even if we add a small delay into the test, users could still trigger the issue in real life | 17:15 |
| johnsom | Yeah, it should stop at the ERROR status | 17:16 |
| johnsom | I can propose a fix to that 404 waiter | 17:18 |
| johnsom | So, address a) | 17:18 |
| frickler | thx, I'll try to reproduce this tomorrow. might actually also be some race condition in pdns | 17:31 |
| *** gmann is now known as gmann_afk | 17:34 | |
| *** gmann_afk is now known as gmann | 18:53 | |
| opendevreview | Merged openstack/designate master: Improve wording for validation error messages https://review.opendev.org/c/openstack/designate/+/820026 | 19:11 |
| eandersson | I noticed that we had tests passing where records took 360 seconds | 20:51 |
| eandersson | and wanted to look into that more | 20:51 |
| opendevreview | Michael Johnson proposed openstack/designate-tempest-plugin master: Abort wait for 404 if the status changes to ERROR https://review.opendev.org/c/openstack/designate-tempest-plugin/+/820075 | 21:54 |
| johnsom | FYI, I have added the review priority tag to the scoped tokens fix patch as we are starting to see other folks run into the issues this patch fixes. I have added a note on the patch with some example links. | 22:52 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!