Add 31 missing airflowctl integration test commands

[steveahnahn]

Problem

Adds integration tests for airflowctl CLI commands. Currently, only 16 of 48 commands are tested.

Solution

Added 31 new command tests (now 47 total):

• Config: +2 (list, lint)
• Connections: +6 (get, update, test, import, delete x2)
• Dags: +4 (get, pause, unpause, update)
• Dag Runs: +2 (get, delete)
• Pools: +7 (get, update, import, export, delete x2)
• Variables: +10 (get, update, import, export, delete x3)

Added 3 JSON fixture files for import command testing (connections, pools, variables).

Commands not included

1. inter-command dependencies: Commands like backfill get/pause/unpause and assets get-queued-events need IDs from prior create operations
2. destructive operations: dags delete
3. state requirements: Asset queue commands require Dags with asset dependencies to be in specific states

Related Issues

• Closes: Add all commands to airflowctl integration tests #58106
• Parent: Add All Commands to airflowctl integration tests #58103
• Related: Add a pre-commit hook to check that airflowctl commands exist in integration tests #58105 (pre-commit hook will need an exclusion list for these intentionally-skipped commands)

[bugraoz93]

Closed and open to trigger full test to run airflowctl integration tests

[bugraoz93]

Looks good! Thanks for the PR. Let's see the CI before merge. Have you run the test with breeze locally?

[steveahnahn]

Good call, I took another look with testing with breeze locally and it seems a lot of them were false positives. The test only checks for specific error strings, so many of the pre-existing tests were found to be using the wrong flags and were silently failing. New commit fixed the wrong CLI flags and positional args, some syntax bugs and ended up disabling a connection test that requires server configs.

Current state now passes correctly. A needed followup is to improve the test_airflowctl_commands.py file now that we're maintaining these airflowctl tests with the upcoming pre-commit pr.

[bugraoz93]

Thanks Steve! Looks great!

[bugraoz93]

Yes, a hook will make things more complete and in syntax.

Exactly we are checking specific errors that states when API returns error which checks if we are doing wrong calls on integration level with endpoints whcih all aising the same error. Additionally we are checking if command is wrong is a way that not with the correct parameters means help text apperas to warn. Maybe we need to add one or more case for false positives