Add HCP status check to run prior to TF commands

[bcmdarroch]

🛠️ Description

Inspired by a recent issue (#172), this PR adds a check to the ConfigureContextFunc hook that pings the HCP status json endpoint (https://pdrzb3d64wsj.statuspage.io/api/v2/components.json) and returns one of the following based on HCP's platform status:

• operational: no error or warning
• outage: error
• degraded_performance or under_maintenance: a warning, which both prints a debug line and only shows if an error is triggered later

NOTE: once hashicorp/terraform-plugin-sdk#790 is fixed, the warnings should show whether or not an error triggers.

This check runs before: terraform plan, terraform apply, terraform refresh, terraform destroy.

$  terraform plan               

│ Error: HCP is currently experiencing an outage. Please check https://status.hashicorp.com/ for more details.
│ 
│   with provider["localhost/providers/hcp"],
│   on example.tf line 15, in provider "hcp":
│   15: provider "hcp" {}
│ 

$  terraform apply

2021-08-10T11:58:19.443-0700 [INFO]  provider.terraform-provider-hcp: 2021/08/10 11:58:19 HCP is undergoing maintenance that may affect performance. Please check https://status.hashicorp.com/ for more details.: timestamp=2021-08-10T11:58:19.443-0700

│ Warning: HCP is undergoing maintenance that may affect performance.
│ 
│   with provider["localhost/providers/hcp"],
│   on example.tf line 15, in provider "hcp":
│   15: provider "hcp" {}
│ 
│ Please check https://status.hashicorp.com/ for more details.
╵
╷
│ Error: something else failed
│ 
│   with provider["localhost/providers/hcp"],
│   on example.tf line 15, in provider "hcp":
│   15: provider "hcp" {}
│ 

🚢 Release Note

Release note for CHANGELOG:

* provider: add HCP status check to run before TF commands [GH-184]

🏗️ Acceptance tests

• Have you run the acceptance tests on this branch?

Output from acceptance testing (just running HVN to do a quick verify):

$ make testacc TESTARGS='-run=TestAccHvn'

--- PASS: TestAccHvn (73.78s)

[xargs-P]

Looks great

[roaks3]

This is really cool! I left some small comments, but approving because the behavior looks good.

[roaks3]

Will this always check the production status? Maybe an edge case, but it seems like this could unnecessarily disrupt testing in lower envs.

[bcmdarroch]

Excellent point! Went ahead and added a HCP_HOST_PATH check so this only runs when pointed at prod.

[roaks3]

Nit: Maybe move this functionality out to a separate file, and then just call something like bool, diags = statuspage.isHCPOperational(). Mainly suggesting this so that the provider/provider.go doesn't become unwieldy as more functionality is added.

[bcmdarroch]

Yeah I like that plan

[tmatilai]

I just saw this in action and I think it's brilliant! 😍
Thanks!

(I mean, the functionality, not the degraded performance 😆)