selectel / terraform-provider-selectel

Terraform Selectel provider

Terraform Selectel Provider

• Website: https://www.terraform.io
• 
• Mailing list: Google Groups

Maintainers

This provider plugin is maintained by:

• Sergei Kolyshkin (@kolsean)
• Nikita Kunets (@dkder3k)
• Ilya Kulakov (@TawR1024)
• Tatyana Voloshina (@T-Sh)

Requirements

• Terraform 0.12.x
• Go 1.21 (to build the provider plugin)

Building The Provider

Clone the repository

$ git clone [email protected]:selectel/terraform-provider-selectel.git

Enter the provider directory and build the provider

$ cd terraform-provider-selectel
$ make build

Using the provider

Getting Started with Terraform at Selectel: kb.selectel.com. Documentation is available at: docs/providers/selectel.
You can find examples in this repository: examples.
Also there are lots of examples in the selectel/terraform-examples.

Developing the Provider

If you wish to work on the provider, you'll first need Go installed on your machine (version 1.17+ is required).

To compile the provider, run make build. This will build the provider and put the provider binary in the current directory.

$ make build

In order to test the provider, you can simply run make test.

$ make test

In order to run the full suite of Acceptance tests, run make testacc.

Note: Acceptance tests create real resources, and often cost money to run.

$ make testacc

Releasing the Provider

This repository contains a GitHub Action configured to automatically build and publish assets for release when a tag is pushed that matches the pattern v* (ie. v0.1.0).

A Gorelaser configuration is provided that produce build artifacts matching the layout required to publish the provider in the Terraform Registry.

Releases will as drafts. Once marked as published on the GitHub Releases page, they will become available via the Terraform Registry.

terraform-provider-selectel's Issues

Support MKS zonal cluster

Library implementation: selectel/mks-go#44

Resell V2 Users resource

We need to support Resell V2 Users API.

MKS Nodegroup V1: fix Create method

We need to return an error in case when nodegroup ID is empty in the Create method.

Support MKS Cluster feature gates & admission controllers

Library implementation: selectel/mks-go#55

Resell V2 Keypairs resource

We need to support Resell V2 Keypairs API.

Upstream reference: go-selvpcclient#110.

Fix an issue with failing Resell V2 Keypair acceptance test

While trying to execute make testacc an error occured:

=== RUN   TestAccVPCV2KeypairImportBasic
    TestAccVPCV2KeypairImportBasic: testing.go:684: Step 1 error: ImportStateVerify attributes not equivalent. Difference is shown below. Top is actual, bottom is expected.
        
        (map[string]string) {
        }
        
        
        (map[string]string) (len=3) {
         (string) (len=9) "regions.#": (string) (len=1) "2",
         (string) (len=18) "regions.1578250041": (string) (len=4) "ru-1",
         (string) (len=18) "regions.1814318523": (string) (len=4) "ru-3"
        }
        
--- FAIL: TestAccVPCV2KeypairImportBasic (92.36s)
FAIL

Add Managed Kubernetes V1 support

This issue tracks the progress on MKS V1 resources.

Add loadbalancers attribute for Resell V2 Floating IP

We must support loadbalancers attribute when it will be available in the go-selvpcclient.

Fix selectel_vpc_crossregion_subnet_v2 project_id

Currently Cross-region subnets retireves their project_id from the first associated subnet.
This should be fixed.

Provider renaming

This issue tracks progress on the provider renaming.

We will rename SelVPC provider to Selectel and all resources will have the following name:

selectel_vpc_subnet_v2
selectel_vpc_project_v2
...

This is needed for supporting other Selectel services apart from VPC.

MKS Cluster V1: uppercase name support

We need to add support for uppercase names in selectel_mks_cluster_v1 resource.
Currently it's kept in lowercase in the terraform state, so if we create a cluster with an uppercase name in the config, the next run of terraform plan command will report diff changes.

Add Selectel DNS API V1 support

Log extended failure messages in case of API errors

We need to log error messages from the API in case of errors.

Fix issues with failing acceptance tests

We often get following errors in the CI acceptance test suite:

------- Stdout: -------
=== RUN   TestAccResellV2SubnetBasic
--- FAIL: TestAccResellV2SubnetBasic (60.03s)
	testing.go:527: Step 0 error: Error applying: 1 error(s) occurred:
		
		* selvpc_resell_project_v2.project_tf_acc_test_1: 1 error(s) occurred:
		
		* selvpc_resell_project_v2.project_tf_acc_test_1: [DEBUG] error creating project: Post https://api.selectel.ru/vpc/resell/v2/projects: net/http: request canceled (Client.Timeout exceeded while awaiting headers)
FAIL

Those kind of issues blocks all other test because we don't have enough limits on the account so we're getting following errors:

------- Stdout: -------
=== RUN   TestAccResellV2ProjectBasic
--- FAIL: TestAccResellV2ProjectBasic (79.76s)
	testing.go:527: Step 3 error: Error applying: 1 error(s) occurred:
		
		* selvpc_resell_project_v2.project_tf_acc_test_1: 1 error(s) occurred:
		
		* selvpc_resell_project_v2.project_tf_acc_test_1: error updating quotas for project '8288493b3be94e54bf9f229291a7938a': selvpcclient: got the 400 error status code from the server
FAIL

Errors: add more context

We need to add more context to error messages.

Fix import of MKS Nodegroup V1 resource

When importing a nodegroup resource, terraform plan suggests updating the value of “nodes_count” in the state, while the correct value is specified in the config.
This should be fixed.

Any plans with adding data resource for domain?

Hi there,
There is resource selectel_domains_record_v1 and I have to create selectel_domains_domain_v1 for domain_id.

Is it hard to add data selectel_domains_domain_v1 to get domain id directly?

Thanks

Unify debug messages

We have several places with duplicate messages like
log.Printf("[DEBUG] Getting keypair %s", d.Id())

We need to use common functions for them to unify debug messages between different resources.

Fix selectel_vpc_keypair_v2 regions

Currently when creating selectel_vpc_keypair_v2 resource without optional regions argument, the next run of terraform plan command will report diff changes.
It should be fixed.

Add support for Kubernetes patch version auto upgrade in MKS Cluster V1

Add Managed Databases V1 support

This issue tracks the progress on DBaaS V1 resources.

Resell V2 Roles resource

We need to support Resell V2 Roles API.

Support MKS Nodegroup autoscaling

Library implementation: selectel/mks-go#53

Allow to import Resell V2 User

We need to implement support of user import when we can retrieve a single user via ID in the go-selvpcclient.

Resell V2 Crossregion networks

We need to add support for the crossregion networks.

Depends on go-selvpcclient#106.

Fix issues with failing acceptance tests

We often get following errors in the CI acceptance test suite:

--- FAIL: TestAccMKSNodegroupV1Basic (596.04s)
13    testing_new.go:22: WARNING: destroy failed, so remote objects may still exist and be subject to billing
14    testing_new.go:22: failed to destroy: terraform failed: exit status 1
15        
16        stderr:
17        
18        Error: error deleting nodegroup 'e3b87082-9265-416c-ba6c-18d0fb951f08/7f67eae2-85c5-4eb8-9a10-25d2804829c8': mks-go: got the 409 status code from the server: {"error":{"message":"Can't perform the operation on the cluster in MAINTENANCE status"}}
19  

Don't use default HTTP client timeouts

We need to use go-selvpcclient with tuned HTTP timeouts.

Resell V2 Token resource

We need to support Resell V2 Tokens API.

Does not update resources

Hello there!
When I'm trying to change project quotas it has no effect

Terraform Version

Terraform v0.12.3

• provider.selectel v2.2.0

Affected Resource(s)

Please list the resources as a list, for example:

• selectel_vpc_project_v2

Terraform Configuration Files

variable "vcpus" {
  default = "4"
}


provider "selectel" {
  token = "${var.sel_token}"
}

resource "selectel_vpc_project_v2" "someproject" {
  name = "someproject"

  quotas {
    resource_name = "compute_cores"
    resource_quotas {
      region = "ru-2"
      zone   = "ru-2b"
      value  = "${var.vcpus}"
    }
  }
}

Expected Behavior

Terraform has to update project quotas

Actual Behavior

Terraform reported that no changes

Acquiring state lock. This may take a few moments...
selectel_vpc_project_v2.someproject: Refreshing state... [id=xxxxxxx]

Apply complete! Resources: 0 added, 0 changed, 0 destroyed.
Releasing state lock. This may take a few moments...

Steps to Reproduce

1. terraform appy: first run
2. terraform apply -var vcpus=6: try to update vcpus number
3. change vcpus variable in tf file and run terraform apply

kubernetes config

Hi, what about getting the kubernetes config via data provider?
like here https://registry.terraform.io/providers/digitalocean/digitalocean/latest/docs/data-sources/kubernetes_cluster

Terraform v0.12

This issue tracks progress on Terraform v0.12 migration.

Add MKS Nodegroup V1 labels support

Use Terraform Plugin SDK v2

https://github.com/hashicorp/terraform-plugin-sdk/blob/master/CHANGELOG.md

Resell V2 Subnet resource

We need to support Resell V2 Subnet API.

Fix an issue with failing Resell V2 Project acceptance test

While trying to execute make testacc an error occurred:

=== RUN   TestAccVPCV2ProjectBasic
    TestAccVPCV2ProjectBasic: testing.go:684: Step 3 error: errors during apply:

        Error: error updating quotas for project 'a1eebfc8ad4f4c31a6d689a22029c88d': selvpcclient: got the 400 status code from the server: {"image_gigabytes":{"new_project_quota":1,"domain_free_quota":0,"project_used_quota":null,"zone":"","region":"ru-1","project_quota":0},"error":"cant_set_quota"}

--- FAIL: TestAccVPCV2ProjectBasic (61.51s)

Update usage and build sections in the documentation

We need to update Using the provider and Building The Provider sections in readme.

Resell V2 VRRP resource

We need to support Resell V2 VRRP API.

Project quotas with empty region or zone don't work

Anyone can get the following errors:

Error: error updating quotas for project 'some-project-id': selvpcclient: got the 500 status code from the server: 500InternalServerErrorServergotitselfintrouble

While using the following config:

quotas {
  resource_name = "network_subnets_29_vrrp"
  resource_quotas {
    value = 1
  }
}

This was fixed in the upstream library https://github.com/selectel/go-selvpcclient/releases/tag/v1.11.1.

Acceptance tests are failing

I guess it started after migrating to the v2 SDK.

First error logs:

cannot run Terraform provider tests: executable found at path /usr/local/bin/terraform is not terraform: located executable at /usr/local/bin/terraform, but output of `terraform version` was:

Your version of Terraform is out of date! The latest version
is 0.13.1. You can update by downloading from https://www.terraform.io/downloads.html
Terraform v0.13.0

After that I added the Terraform installation step into acceptance tests (#132) and disabled debug logging (#134) and now it's failing with logs:

cannot run Terraform provider tests: executable found at path /home/runner/work/_temp/839df762-2d41-4108-815e-916ae2727fec/terraform is not terraform: located executable at /home/runner/work/_temp/839df762-2d41-4108-815e-916ae2727fec/terraform, but output of `terraform version` was:
/home/runner/work/_temp/839df762-2d41-4108-815e-916ae2727fec/terraform-bin version
Terraform v0.13.2

Support MKS Kubernetes options

Library implementation: selectel/mks-go#47

Add support for Kubernetes minor version upgrade in MKS Cluster V1

Fix an issue with failing Resell V2 Token acceptance test

While trying to execute make testacc an error occured:

=== RUN   TestAccResellV2TokenAccount
--- FAIL: TestAccResellV2TokenAccount (0.23s)
    testing.go:527: Step 0 error: Error applying: 1 error(s) occurred:
        
        * selvpc_resell_token_v2.token_tf_acc_test_1: 1 error(s) occurred:
        
        * selvpc_resell_token_v2.token_tf_acc_test_1: error creating token: selvpcclient: got the 401 error status code from the server
FAIL

Support DBaaS configuration parameters

Library implementation: selectel/dbaas-go#6

Use GolangCI-Lint in Travis CI

We need to use GolangCI-Lint with linters and remove separate scripts for errs, vet, critic, etc.

Support MKS Nodegroup taints

Library implementation: selectel/mks-go#50

Retries on 5XX from Domains API

From time to time requests to Domains API fail with 5xx HTTP status codes. For the sake of durability it makes sense to do retries when Domains API returns 5xx HTTP status codes.

This library looks like a good solution:
https://github.com/hashicorp/go-retryablehttp

Allow to delete Resell V2 Token

We need to implement support of token deletion when it will be available in the go-selvpcclient.

Add VPC V2 License network and subnet data

We should support network_id and subnet_id in selectel_vpc_license_v2 resource.

[PROPOSAL] Switch to Go Modules

As part of the preparation for Terraform v0.12, we would like to migrate all providers to use Go Modules. We plan to continue checking dependencies into vendor/ to remain compatible with existing tooling/CI for a period of time, however go modules will be used for management. Go Modules is the official solution for the go programming language, we understand some providers might not want this change yet, however we encourage providers to begin looking towards the switch as this is how we will be managing all Go projects in the future. Would maintainers please react with 👍 for support, or 👎 if you wish to have this provider omitted from the first wave of pull requests. If your provider is in support, we would ask that you avoid merging any pull requests that mutate the dependencies while the Go Modules PR is open (in fact a total codefreeze would be even more helpful), otherwise we will need to close that PR and re-run go mod init. Once merged, dependencies can be added or updated as follows:

$ GO111MODULE=on go get github.com/some/module@master
$ GO111MODULE=on go mod tidy
$ GO111MODULE=on go mod vendor

GO111MODULE=on might be unnecessary depending on your environment, this example will fetch a module @ master and record it in your project's go.mod and go.sum files. It's a good idea to tidy up afterward and then copy the dependencies into vendor/. To remove dependencies from your project, simply remove all usage from your codebase and run:

$ GO111MODULE=on go mody tidy
$ GO111MODULE=on go mod vendor

Thank you sincerely for all your time, contributions, and cooperation!

MKS V1: increase default timeouts

We need to increase default timeouts for selectel_mks_cluster_v1 and selectel_mks_nodegroup_v1 resources to avoid errors during long time operations.