| frickler | jamespage_: same question as knikolla, why wouldn't different regions work? other than that, it isn't currently possible afaict | 07:55 |
|---|---|---|
| *** whoami-rajat__ is now known as whoami-rajat | 11:33 | |
| *** dviroel|out is now known as dviroel | 13:00 | |
| knikolla | dmendiza[m], d34dh0r53: BU is observing Juneteenth today rather than Monday, so I won't be around for the reviewathon. | 13:34 |
| opendevreview | Stephen Finucane proposed openstack/keystone master: tox: Don't generate byte code https://review.opendev.org/c/openstack/keystone/+/846365 | 13:38 |
| opendevreview | Stephen Finucane proposed openstack/keystone master: Ignore UserWarning for scope checks during test runs https://review.opendev.org/c/openstack/keystone/+/846366 | 13:38 |
| opendevreview | Stephen Finucane proposed openstack/keystone master: sql: Don't create a new connection in migrations https://review.opendev.org/c/openstack/keystone/+/846367 | 13:38 |
| opendevreview | Stephen Finucane proposed openstack/keystone master: WIP sql: Don't rely on globals to configure enginefacades https://review.opendev.org/c/openstack/keystone/+/846368 | 13:38 |
| opendevreview | Stephen Finucane proposed openstack/keystone master: sql: Integrate alembic https://review.opendev.org/c/openstack/keystone/+/825844 | 13:38 |
| opendevreview | Stephen Finucane proposed openstack/keystone master: docs: Update docs to reflect migration to Alembic https://review.opendev.org/c/openstack/keystone/+/828905 | 13:39 |
| opendevreview | Stephen Finucane proposed openstack/keystone master: WIP: sql: Add support for auto-generation https://review.opendev.org/c/openstack/keystone/+/826147 | 13:39 |
| *** dviroel is now known as dviroel|pto | 14:03 | |
| *** whoami-rajat__ is now known as whoami-rajat | 14:43 | |
| dmendiza[m] | knikolla: no worries, thanks for the heads up. | 15:14 |
| stephenfin | zzzeek: At some point I'm going to need your eyes on https://review.opendev.org/c/openstack/keystone/+/825844 again. I can't figure out what's going wrong. *Something* is causing multiple tests to use the same DB connection and trample on each other, but I haven't a clue what and Keystone's session management is totally different to cinder and nova's :-( | 16:26 |
| zzzeek | stephenfin: OK. I will try to find time, did you ping me on the gerrit so i have it in my email? | 16:26 |
| zzzeek | there's an immense amount of things people need me to look at | 16:26 |
| stephenfin | I did a while back. Just removed and re-added you | 16:27 |
| stephenfin | That's fair and I'm reluctant to even ask, but this is maybe the fourth or fifth day (not in a row) I've burnt trying to untangle this and I'm getting nowhere fast /o\ My call for help at the Summit didn't yield results either unfortunately :) | 16:28 |
| zzzeek | stephenfin: I think i might have seen the email yesterday? or was this longer ago? | 16:31 |
| stephenfin | Oh, like 4 months ago | 16:31 |
| zzzeek | omg sorry, that i missed | 16:31 |
| zzzeek | stephenfin: is there a way for me to run the keystone tests in question without running the whole suite and still reproduce the issue? | 16:32 |
| stephenfin | not that I've found, which is what makes it such a pain. I tried running stestr with the '--analyze-isolation' flag to see if I could identify the interactions but to no avail | 16:33 |
| stephenfin | if I could find some way to differentiate between different in-memory SQLite instances (so I could say "oh, these two tests are using the same instance") that would probably do the trick but I haven't figured out a way to do that | 16:34 |
| zzzeek | ok any help from keystone devs? | 16:35 |
| stephenfin | I don't know if there are many keystone devs left | 16:35 |
| stephenfin | other than dmendiza[m] perhaps | 16:35 |
| zzzeek | stephenfin: oh geez I thought keystone was on alembic already, uh | 16:41 |
| zzzeek | wow | 16:41 |
| zzzeek | stephenfin: this is huge | 16:42 |
| zzzeek | stephenfin: what happens if you just run keystone migrations from CLI using this? does that work? | 16:42 |
| stephenfin | yeah, and the tests running in serial order work just fine | 16:43 |
| stephenfin | I've tried to align keystone with nova, namely, using a global enginefacade and monkeypatching it on a per-test basis but that's borked and I don't know why https://review.opendev.org/c/openstack/keystone/+/846368 | 16:44 |
| stephenfin | (as a way of eliminating shared global state) | 16:44 |
| zzzeek | oh, tests running seril is fine, and using ...distinct proceses? they fail? | 16:44 |
| zzzeek | test suite doesnt use threads right | 16:45 |
| stephenfin | yeah, stestr uses processes afaik | 16:45 |
| zzzeek | stephenfin: OK! and tests fail w sqlite? | 16:47 |
| stephenfin | and MySQL and PostgreSQL | 16:47 |
| zzzeek | ok and ...is each process using a separate database? | 16:47 |
| stephenfin | It should be, but I don't think it is | 16:47 |
| zzzeek | OK but this is jusit a top level "use the right database" issue then , within the test suites | 16:48 |
| stephenfin | Because the errors are usually "this table is missing" | 16:48 |
| zzzeek | sure | 16:48 |
| opendevreview | mike_mp@zzzcomputing.com proposed openstack/keystone master: sql: Integrate alembic https://review.opendev.org/c/openstack/keystone/+/825844 | 18:08 |
| zzzeek | stephenfin: it's likely I found the issue, will see | 18:08 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!