azurerm_frontdoor, azurerm_linux_virtual_machine, azurerm_windows_virtual_machine, azurerm_ssh_public_key - Fix nil panic by throwing error when the resource was missing during update

[lonegunmanb]

When executing update, if the resource was deleted out-of-band already, the current implementation would lead to a nil panic. This pr fix #21972 throwing the error even http status code is 404.

Another possible approach is like what we do in azurerm_linux_virtual_machine, just swallow the error and do nothing in this scenario.

I personally prefer throwing out the error since the real situation is different than the operator's assumption, the provider should notify the user about this drift explicitly.

[tombuildsstuff]

hey @lonegunmanb

Thanks for this PR.

Taking a look through here the usage that you've called out within the Linux (and presumably Windows) Virtual Machine resources is actually a bug, so it'd be worth removing that as a part of this PR too - would you mind looking into fixing that? From Terraform's perspective to get into the Update function we'll have passed through the Read, where we'll have determined that the resource has gone (via d.SetID("")) - as such the update should be able to assume that this resource exists, and throw the error if it's not found?

Thanks!

[lonegunmanb]

Glad to do so @tombuildsstuff!

---

I've updated the pr and scanned compute package, more similar bugs have been fixed in the same way, I think compute and frontdoor package don't contain this bug any more, but maybe we need arrange a check on other package, or a static check tool based on go ast to detect such issue. @tombuildsstuff

[lonegunmanb]

Hi @tombuildsstuff, a kindly ping, could we merge this pr? Thanks!

[stephybun]

Thanks for this @lonegunmanb LGTM 🥝

[github-actions]

I'm going to lock this pull request because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active contributions.
If you have found a problem that seems related to this change, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.