K8s Migration Notes

# [Kubernetes Service Migration Prep Checklist](https://subsplash.atlassian.net/wiki/spaces/SRE/pages/268665112/How+to+migrate+a+service+to+Kubernetes#Code-Migration-Checklist)

## Rectify json configuration file template

1. Download the template for the service, put it in config/, and add it to the repo.
 - Copy from (S3 bucket) 00-config/templates/
   - $objectPath = 'templates/{environment}/{service}/{servicename}.json.tmpl'
     - Example: templates/stage/push/push.json.tmpl
   - $templateList = Get-S3Object -BucketName 00-config | Where-Object -FilterScript {$_.Key -like 'templates/*' }
   - $template = Get-S3Object -BucketName 00-config -Key $objectPath

SEE https://subsplash.io/go/kubernetes-example-app

## Setup Helm Chart & Values

```go
gen chart <your-service> -port <port#>  (see port for service from: https://subsplash.atlassian.net/wiki/x/34JBDg)
```

## Dockerfile & publish-image job

In the .gitlab-ci.yaml file add the following jobs if not already there

```yaml
publish-image:
  extends: .publish-image
  variables:
    PORT: # Set port for service from https://subsplash.atlassian.net/wiki/x/34JBDg

publish-db-image:
  extends: .publish-db-image
gitlab-ci.yaml pipeline & k8s deployment jobs

# Add CHART_VERSION {latest version} here 

variables:
  CHART_VERSION: 1.x # https://subsplash.io/ops/kubernetes-app-helm-chart
Add deploy-eks:* stages (note once the deployment works switch the deploy to.deploy-ec2-eks:* ) see How debug eks deployments for more information.

deploy:sandbox:
  extends: .deploy-ec2-eks:sandbox

deploy:dev:
  extends: .deploy-ec2-eks:dev

deploy:stage:
  extends: .deploy-ec2-eks:stage 

deploy:prod:
  extends: .deploy-ec2-eks:prod

```

## Assume Role Support for dependency on AWS resources (Only add if service requires AWS resources, skip otherwise)

Add the required AWS libraries to app.go or main.go. At minimum you will need:

```go
import (      
        "github.com/aws/aws-sdk-go/aws/credentials/stscreds"
        "github.com/aws/aws-sdk-go/aws/session"
)
Add this to the bottom of app.go (or main.go if app.go doesn’t exist):

// setAssumeRoleCreds sets the credentials to allow assuming
// a role based on the AWS_ROLE_ARN env var being set.
func setAssumeRoleCreds(s *session.Session, c *aws.Config) {
	if roleARN := os.Getenv("AWS_ROLE_ARN"); roleARN != "" {
		c.Credentials = stscreds.NewWebIdentityCredentials(
			s,
			roleARN,
			os.Getenv("POD_NAME"),
			os.Getenv("AWS_WEB_IDENTITY_TOKEN_FILE"),
		)
	}
}
```

```go
Put this before setting up clients for AWS services:

		snsConf := &aws.Config{
			Region: aws.String("us-west-2"),
		}
		if conf.App.Env == service.EnvDocker {
			snsConf.WithCredentials(credentials.NewStaticCredentials("foo", "bar", ""))
		}
		if e := os.Getenv("SNS_ENDPOINT"); e != "" {
			snsConf.WithEndpoint(e)
		}

		snsSession, err := session.NewSession(snsConf)
		if err != nil {
			log.Fatal("error initializing an aws session: ", err)
		}

		setAssumeRoleCreds(snsSession, snsConf)
```

## Container aware cpu limits for go services

Add "go.uber.org/automaxprocs/maxprocs" to the file imports.

Add "strconv" to file imports.

Put this at the beginning of app.New(), or app.Run() if that doesn’t exist

```go
if isK8S, _ := strconv.ParseBool(os.Getenv("KUBERNETES")); isK8S {
    _, err := maxprocs.Set(maxprocs.Logger(log.Printf))
    if err != nil {
        log.Fatalf("failed to set GOMAXPROCS: %v", err)
    }
}
```
Update go modules & vendored files as needed.