fix: allow orchestrator_version if auto-upgrade is 'patch' to allow default_node_pool upgrade

[aescrob]

Describe your changes

-  # - patch, but then the kubernetes_version must not specify a patch number and orchestrator_version must be null
+  # - patch, but then neither the kubernetes_version nor orchestrator_version must specify a patch number

Issue number

#301

Checklist before requesting a review

• The pr title can be used to describe what this pr did in CHANGELOG.md file
• I have executed pre-commit on my machine
• I have passed pr-check on my machine

Thanks for your cooperation!

[the-technat]

LGTM

[lonegunmanb]

Almost LGTM if the e2e test could pass, but would you please update var.automatic_channel_upgrade's description about this new requirement for orchestrator_version when var.automatic_channel_upgrade = patch? Thanks @aescrob

[lonegunmanb]

Thanks @aescrob for updating this pr, we have some e2e tests failed, one example is:

TestExamplesStartup 2023-02-16T02:25:55Z retry.go:99: Returning due to fatal error: FatalError{Underlying: error while running command: exit status 1; ╷
│ Error: Resource precondition failed
│ 
│   on ../../main.tf line 296, in resource "azurerm_kubernetes_cluster" "main":
│  296:       condition     = local.automatic_channel_upgrade_check
│     ├────────────────
│     │ local.automatic_channel_upgrade_check is false
│ 
│ Either disable automatic upgrades, or only specify up to the minor version
│ when using `automatic_channel_upgrade=patch` or don't specify
│ `kubernetes_version` at all when using
│ `automatic_channel_upgrade=stable|rapid|node-image`. With automatic
│ upgrades `orchestrator_version` must be set to `null`.
╵}
=== CONT  TestExamplesStartup
    apply.go:34: 
        	Error Trace:	/src/test/e2e/apply.go:34
        	            				/src/test/e2e/e2etest.go:57
        	            				/src/test/e2e/e2etest.go:38
        	            				/src/test/e2e/terraform_aks_test.go:22
        	Error:      	Received unexpected error:
        	            	FatalError{Underlying: error while running command: exit status 1; ╷
        	            	│ Error: Resource precondition failed
        	            	│ 
        	            	│   on ../../main.tf line 296, in resource "azurerm_kubernetes_cluster" "main":
        	            	│  296:       condition     = local.automatic_channel_upgrade_check
        	            	│     ├────────────────
        	            	│     │ local.automatic_channel_upgrade_check is false
        	            	│ 
        	            	│ Either disable automatic upgrades, or only specify up to the minor version
        	            	│ when using `automatic_channel_upgrade=patch` or don't specify
        	            	│ `kubernetes_version` at all when using
        	            	│ `automatic_channel_upgrade=stable|rapid|node-image`. With automatic
        	            	│ upgrades `orchestrator_version` must be set to `null`.
        	            	╵}
        	Test:       	TestExamplesStartup

Would you please fix our examples and tests, then verify them by run e2e tests or just run terraform plan for all examples before you commit? Thanks for your understanding!

Btw, in case this pr failed the version update test, which means when the previous version of the example would met errors when it upgrades to this pr version, then we must consider this pr as a breaking change. In that case, I would rather postpone the merge, and release our current version as a new minor version so our users don't need to take the risks to get current unreleased features.

[aescrob]

Hi @lonegunmanb - thanks for your support.
I've changed the precondition to still allow orchestrator_version==null. Like this it is not a breaking change and the examples are not required to be fixed.

[lonegunmanb]

LGTM but I'd like to ask @the-technat for a review.

[the-technat]

Makes sense to me and looks good. Maybe it's worth documenting that the orchestrator_version can be set to null because it should be a non-breaking change and not because it's a recommended approach to configure the module.

[lonegunmanb]

Thanks @aescrob, LGTM! 🚀