Upgrade to use API version 2016-04-01

[tombuildsstuff]

Original Issue: coreos/tectonic-installer#1315 / Original PR: jen20/riviera#24

It appears that Azure's gradually removing support for the version of the DNS API that we're currently using via the Riviera SDK 2015-05-04-preview. Whilst we could switch over to using version 2016-04-01 it appears this requires some further refactoring where fields have changed.

The following PR's have been opened recently to switch these resources over to the Azure SDK for Go - which makes use of the updated API:

• resource_arm_dns_zone: switch dependency from riviera to azure-sdk-for-go #188
• resource_arm_dns_txt_record: switch dependency from riviera to azure-sdk-for-go #187
• resource_arm_dns_srv_record: switch dependency from riviera to azure-sdk-for-go #186
• azurerm_dns_ptr_record - refactoring to match the DNS updates #195
• resource_arm_dns_mx_record: switch dependency from riviera to azure-sdk-for-go #185
• Resource arm dns cname record #182
• resource_arm_dns_aaaa_record: switch dependency from riviera to azure-sdk-for-go #181
• resource_arm_dns_a_record: switch dependency from riviera to azure-sdk-for-go #180
• resource_arm_dns_ns_record: switch dependency from riviera to azure-sdk-for-go #179

Given these PR's exist - it looks like it could be quicker to switch these over to the Azure SDK for Go

[justaugustus]

@tombuildsstuff Thanks for your attention on this! Agreed that switching to Azure SDK for Go feels like the right route.

(Adding @alexsomesan, @s-urbaniak, @metral, @robszumski for visibility on the CoreOS side.)

[discordianfish]

That's a huge issue. This effectively means everyone managing their azure setup with terraform are unable to do any (DNS) changes to their infrastructure right now. I don't even want to imagine what people going through right now who run prod stuff managed with terraform on azure and are effectively locked out of their infrastructure.
Has anyone contacted Azure to see if they could re-enable the APIs again?

[jdumars]

I'm looking into this now.

[tombuildsstuff]

Hey @discordianfish

Has anyone contacted Azure to see if they could re-enable the APIs again?

That's a really good point - I've raised this internally at our end too and I'll post here when we've got an update (thanks @jdumars for requesting too)

That's a huge issue. This effectively means everyone managing their azure setup with terraform are unable to do any (DNS) changes to their infrastructure right now. I don't even want to imagine what people going through right now who run prod stuff managed with terraform on azure and are effectively locked out of their infrastructure.

Apologies for the inconvenience here - I completely agree this is a major issue and we’re looking into it right now. Unfortunately the removal of the API version appears to be rolling out gradually and we were unaware this version was due to be removed. Since we were notified of the issue we’ve been working through upgrading to the new API version - and that should be completed in the next couple of hours. We’ll also be double-checking the versions of the API’s we’re using as a result of this for any further deprecations.

Apologies once again for the inconvenience - as I mentioned above we should have a fix for this completed later today.

Thanks!

[jdumars]

Waiting to confirm on this, but I believe the issue has been fixed in Terraform master, and will be closed in the next update.

[tombuildsstuff]

Hey @justaugustus @discordianfish @jdumars

Waiting to confirm on this, but I believe the issue has been fixed in Terraform master, and will be closed in the next update.

Correct - I was just running the entire test suite before releasing :)

I've just released version 0.1.4 of the Terraform AzureRM Provider which switches over to using version 2016-04-01 of the Azure DNS API's - which is available in Terraform 0.10-rc1 (and can be used in Terraform 0.9 by updating the ~/.terraformrc file to reference the downloaded provider).

Would it be possible for you to take a look? :)

Thanks!

cc @alexsomesan, @s-urbaniak, @metral, @robszumski

[justaugustus]

Thank you so much for chasing this down, @tombuildsstuff!
CoreOS team will give this a try and report back when we have an opportunity.

[alexsomesan]

@justaugustus I will test this out tomorrow

[tombuildsstuff]

Given the fix for this issue has been released - I'm going to close this issue for the moment. However please feel free to re-open it if there's still an issue here :)

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. If you feel I made an error 🤖 🙉 , please reach out to my human friends 👉 hashibot-feedback@hashicorp.com. Thanks!