| *** spatel has joined #openstack-trove | 01:03 | |
| sorrison | Hi lxkong, I'm having a play with tempest in the gate and I'm getting build timeouts `status(current: BUILD) within the required time (196 s).` | 01:15 |
|---|---|---|
| sorrison | Can't figure out why the timeout is so low | 01:16 |
| sorrison | ahh, of course now that I ask I figure it out | 01:17 |
| *** openstackgerrit has joined #openstack-trove | 01:20 | |
| openstackgerrit | Sam Morrison proposed openstack/trove-tempest-plugin master: [WIP] reinstate removed api tests https://review.opendev.org/c/openstack/trove-tempest-plugin/+/761472 | 01:20 |
| *** spatel has quit IRC | 01:55 | |
| *** rcernin has quit IRC | 02:19 | |
| *** rcernin has joined #openstack-trove | 02:21 | |
| *** rcernin has quit IRC | 02:23 | |
| *** rcernin has joined #openstack-trove | 02:24 | |
| *** spatel has joined #openstack-trove | 02:33 | |
| *** spatel has quit IRC | 02:37 | |
| *** sapd1 has joined #openstack-trove | 04:17 | |
| lxkong | sorrison: hi, still need help? | 04:54 |
| sorrison | hey lxkong | 04:54 |
| lxkong | i'm preparing the trove upgrade recently so may not pay much attention to the irc | 04:54 |
| sorrison | been trying to get the tempest tests working in the gate with that nested virt nodegroup | 04:55 |
| sorrison | https://review.opendev.org/c/openstack/trove/+/764073 should fix the devstack issue | 04:55 |
| lxkong | after jenkins +1 the other day, i can't make it work :-( | 04:56 |
| sorrison | That patch will fix your issue | 04:56 |
| lxkong | i will rebase mine and see if that could pass | 04:56 |
| sorrison | I got a full devstack run to work ok with https://review.opendev.org/c/openstack/trove-tempest-plugin/+/761472 which depends on that | 04:57 |
| openstackgerrit | Lingxian Kong proposed openstack/trove master: Use nested virt node for trove functional test https://review.opendev.org/c/openstack/trove/+/763630 | 04:58 |
| sorrison | Although the tempest tests fail to spin up a instance `Trove instance status: ERROR, Trove instance status description: Build error: guestagent timeout.: trove.common.exception.PollTimeOut: Polling request timed out.` | 04:58 |
| sorrison | I see that in the taskmanager logs | 04:58 |
| sorrison | do we get the guestagent logs in zuul somehow do you know? | 04:59 |
| lxkong | in tempest, no, only supported in trove-scenario-mysql-single | 05:00 |
| sorrison | ok, I'll look at how that is configured | 05:02 |
| lxkong | sorrison: may i know what are you working on when you saw this issue? | 05:02 |
| sorrison | I'm trying to add back all the API tempest tests https://review.opendev.org/c/openstack/trove-tempest-plugin/+/761472 | 05:02 |
| lxkong | hmm..i would suggest we compare what's missing with the current test cases and add those one by one, instead of restoring the previous code. | 05:04 |
| lxkong | e.g. we really don't need those xxx_client.py | 05:04 |
| sorrison | ah ok, why not? | 05:05 |
| sorrison | the separate clients in tempest are really important to ensure the api doesn't break | 05:06 |
| sorrison | that's how it's done in the other projects so be good to keep it consistent | 05:06 |
| lxkong | the methods in trove_tempest_plugin/services/client.py can cover most of the api calls, no need to add extra wrapper | 05:06 |
| lxkong | but i'm not against to have those, just suggestion | 05:07 |
| sorrison | I just think doing it that way make it easy to introduce backwards incompatible changes | 05:07 |
| lxkong | also, most of the test cases in trove_tempest_plugin/tests/api/database/instances/test_instance_actions.py have been covered in the current repo | 05:08 |
| lxkong | and the instance actions, user and db mgmt, backup are mixed in that file | 05:09 |
| lxkong | and no consideration for different datastores | 05:09 |
| sorrison | Yeah the API tests don't test the specific datastore, that doesn't matter | 05:10 |
| lxkong | e.g. we need to be able to test not only mysql, but also mariadb, and postgresql as well | 05:10 |
| sorrison | it just tests the API | 05:10 |
| sorrison | but yes the API tests work on what ever backend you like | 05:10 |
| sorrison | would also be good to bring it back more inline with our tempest tests are written, makes it easier for people like me who work on lots of openstack projects | 05:13 |
| lxkong | yeah, if it's me, i will just add missing test cases based on the current trove tempest structure to avoid duplication | 05:16 |
| lxkong | something we could improve | 05:16 |
| sorrison | ok, I'll have at the scenario tests and see where the duplicates are an update this patch | 05:17 |
| lxkong | the current tempest, when you run `tempest run --list-tests --regex ^trove_tempest_plugin` | 05:17 |
| lxkong | you can see test cases for different datastore | 05:17 |
| lxkong | but for those api tests, it'd hard coded to use `CONF.database.datastore_type` | 05:18 |
| lxkong | it's not ideal when you run tempest, you need to change this option frequently for different datastores | 05:18 |
| lxkong | although they are all api tests, but we still want to make sure the api does work for all the datastores that are supported. | 05:19 |
| lxkong | because same api calls could invoke the same methods of different datastores | 05:20 |
| lxkong | make sense? | 05:20 |
| sorrison | we just have a tempest env for each datastore we support | 05:20 |
| sorrison | it looks like you're doing that in the gate too as it is called `trove-scenario-mysql-single` ? | 05:21 |
| lxkong | this one is a legacy one | 05:22 |
| sorrison | so I thought you'd have a `trove-scenario-postgresql-single` | 05:22 |
| sorrison | ah ok | 05:22 |
| lxkong | the legacy ones are maintained but shouldn't add more test cases there | 05:22 |
| sorrison | it might make it less complicated if the devstack plugin and tempest were set up to assume one datastore? | 05:23 |
| sorrison | it would make it faster to run too? | 05:23 |
| sorrison | then can run each datastore in parallel | 05:24 |
| lxkong | for upstream CI, yes, we can set up multiple jobs to run different ds, but tempest itself should support multiple datastores in an easy way | 05:24 |
| lxkong | just like how the scenario test is implemented currently | 05:24 |
| sorrison | yeah ok, it does add a lot of complication though | 05:25 |
| lxkong | because trove is kind of, special | 05:25 |
| sorrison | if you go away then it is a lot of work to maintain all of this | 05:25 |
| lxkong | it's not that complicated, one abstract layer and different layers for different datastores | 05:26 |
| lxkong | no duplication at all | 05:26 |
| lxkong | if you want to run tests for mysql, just `tempest run --list-tests --regex ^trove_tempest_plugin | grep -i mysql` | 05:27 |
| *** sapd1 has quit IRC | 05:27 | |
| sorrison | yeah that's all good | 05:27 |
| lxkong | it's also explicitly tell the developers, what datastores are supported in the upstream | 05:27 |
| lxkong | or tested | 05:28 |
| lxkong | ok, i will continue with my upgrade work | 05:28 |
| lxkong | hopefully we could finish the upgrade in 2020 | 05:29 |
| sorrison | what are you upgrading? | 05:29 |
| lxkong | trove from ussuri to victoia | 05:29 |
| lxkong | but there are also other services involved | 05:29 |
| sorrison | yeah ok, we are going to be stuck on ussuri for a while I think until all the container stuff with upgrades is sorted out | 05:30 |
| lxkong | yeah, i know it's painful, but i believe we could work that out in a sensible way together. | 05:32 |
| sorrison | I fear that the changes are so great that it's going to be very hard for us | 05:33 |
| lxkong | heh, we will see | 05:33 |
| * lxkong really hopes he could travel back to 2 years ago so he could wok with sorrison together on trove, rather than by himself | 05:35 | |
| sorrison | yeah I feel for you :-) | 05:35 |
| sorrison | good luck with the upgrade | 05:36 |
| lxkong | ty | 05:36 |
| sorrison | The hard thing is that most deployments are a couple releases behind so we don't see these changes until it is too late | 05:37 |
| lxkong | yeah it's disappointed to know there were actually many other trove deployments out there but no one showed up in the community | 05:38 |
| lxkong | during the hardest time of this project | 05:38 |
| lxkong | i know there are some other clouds have deployed trove, but have added many private changes and decided to not follow the upstream | 05:39 |
| lxkong | that's totally understandable | 05:39 |
| sorrison | yeah it's tricky | 05:39 |
| lxkong | that's also part of open source | 05:40 |
| * lxkong has to go for dinner | 05:40 | |
| *** sapd1 has joined #openstack-trove | 06:05 | |
| *** spatel has joined #openstack-trove | 08:25 | |
| *** spatel has quit IRC | 08:30 | |
| *** rcernin has quit IRC | 08:50 | |
| *** tosky has joined #openstack-trove | 08:50 | |
| *** rcernin has joined #openstack-trove | 09:18 | |
| *** rcernin has quit IRC | 09:27 | |
| *** rcernin has joined #openstack-trove | 09:37 | |
| *** sapd1 has quit IRC | 10:00 | |
| *** rcernin has quit IRC | 10:07 | |
| *** rcernin has joined #openstack-trove | 10:27 | |
| *** rcernin has quit IRC | 10:33 | |
| *** rcernin has joined #openstack-trove | 10:59 | |
| *** rcernin has quit IRC | 11:04 | |
| *** spatel has joined #openstack-trove | 11:26 | |
| *** spatel has quit IRC | 11:31 | |
| *** e0ne has joined #openstack-trove | 11:38 | |
| *** rcernin has joined #openstack-trove | 11:47 | |
| *** rcernin has quit IRC | 12:07 | |
| *** rcernin has joined #openstack-trove | 13:21 | |
| *** e0ne has quit IRC | 13:25 | |
| *** e0ne has joined #openstack-trove | 13:25 | |
| *** rcernin has quit IRC | 13:25 | |
| *** congnt96 has joined #openstack-trove | 14:37 | |
| *** sapd1 has joined #openstack-trove | 14:58 | |
| *** irclogbot_0 has quit IRC | 16:41 | |
| *** irclogbot_3 has joined #openstack-trove | 16:47 | |
| *** e0ne has quit IRC | 18:42 | |
| *** congnt96 has quit IRC | 21:03 | |
| *** sapd1 has quit IRC | 21:05 | |
| *** sapd1 has joined #openstack-trove | 21:17 | |
| *** e0ne has joined #openstack-trove | 21:30 | |
| *** e0ne has quit IRC | 21:31 | |
| *** rcernin has joined #openstack-trove | 21:35 | |
| *** rcernin has quit IRC | 21:37 | |
| *** rcernin has joined #openstack-trove | 21:38 | |
| *** e0ne has joined #openstack-trove | 21:46 | |
| *** e0ne has quit IRC | 21:52 | |
| openstackgerrit | Lingxian Kong proposed openstack/trove master: Support --image-tags for trove-manager datastore_version_update command https://review.opendev.org/c/openstack/trove/+/764245 | 22:09 |
| *** rcernin has quit IRC | 22:54 | |
| *** rcernin has joined #openstack-trove | 22:54 | |
| openstackgerrit | Merged openstack/trove master: Support --image-tags for trove-manager datastore_version_update command https://review.opendev.org/c/openstack/trove/+/764245 | 23:50 |
| *** tosky has quit IRC | 23:51 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!