tests: stop using set_keyspace in test_can_register_udt_before_conn…

[fruch]

…ecting

this test is doing multiple is using USE comamnd, seems like we are having same race conditions with the applying of those, and getting the following error once in a while:

>   raise self._final_exception
E   cassandra.InvalidRequest: Error from server: code=2200 [Invalid query] message="Unknown field 'is_cool' in value of user defined type user"

in this test we'll use full qulified names of the table with the keyspace and not set_keyspsce (i.e. USE command)

Fixes: https://github.com/scylladb/python-driver/issues/264

[fruch]

Let take it for an actual run in CI (where the flakiness usually surface)

I'm still not total sure it's just a schema mismatch, and not issue with the handling of set_keyspace inside the driver. maybe similar to #190

[fruch]

CI is failing cause missing #300, we'll merge that and then give this one a run

[Lorak-mmk]

CI is failing cause missing #300, we'll merge that and then give this one a run

I merged #300

Let take it for an actual run in CI (where the flakiness usually surface)

I'm still not total sure it's just a schema mismatch, and not issue with the handling of set_keyspace inside the driver. maybe similar to #190

I wouldn't be surprised, I remember that set_keyspace handling was complicated (to say the least) in this driver.

[fruch]

@Lorak-mmk

I've changed this to stop using set_keyspace, that not the goal of that specific test, and seems like the source of it's instability

[Lorak-mmk]

Ok, then let's merge it, but first could you open an issue about set_keyspace being flaky and attach original code of this test there?

[fruch]

Ok, then let's merge it, but first could you open an issue about set_keyspace being flaky and attach original code of this test there?

raised the issue: https://github.com/scylladb/python-driver/issues/302