kit-deployer

kit-deployer

Use this service to deploy files to multiple Kubernetes clusters. You just have to organize your manifest files into directories that match the names of your clusters (the name defined in your kubeconfig files). Then you just provide a directory of kubeconfig files and the kit-deployer will asynchronously send all manifests up to their corresponding clusters.

There is also support for namespaces. Simply provide a directory with namespaces that are grouped into folders that match the name of the kubeconfig cluster you want them deployed to.

The kit-deployer service was designed to work for a CI type service like Codeship where it can be run as a docker image as part of your automated workflow. However, you can also use it as an npm module or as a CLI tool directly.

Use as Docker Image

docker run quay.io/invision/kit-deployer --help

We recommend using kit components with Codeship's Docker Infrastructure, however you are free to run this tool however way you wish. Anything that has Docker can run this image.

Using as CLI

You can run the ./src/deployer --help to see how it works.

Note this method requires node and was tested on version 5.5.0.

Using as npm module

Use npm to install kit-deployer:

$ npm install kit-deployer --save

Then require it and use it like so:

var Deployer = require("kit-deployer").Deployer;

var options = {
	sha: "c6350f4c2709708b8f784408a440030e704b2b9a",
	dryRun: true,
	diff: true,
	github: {
		token: "ccd50691c75bc7bae0a5490ea08ff9dcc9c264a5",
		user: "chesleybrown",
		repo: "my-app"
	}
};

var deployer = new Deployer(options);

// setup logging of events
deployer.on("info", function(message) {
	console.log(message);
});
deployer.on("warn", function(message) {
	console.log(message);
});
deployer.on("error", function(message) {
	console.log(message);
});
deployer.on("fatal", function(message) {
	console.log(message);
});
deployer.on("status", function(status) {
	console.log(status);
});
deployer.on("progress", function(progress) {
	console.log(progress);
});

deployer
	.deploy("/configs/**/kubeconfig", "/manifests", "/namespaces")
	.then(console.log)
	.catch(console.error)
	.done();

Note this method requires node and was tested on version 5.5.0.

Progress

The progress event emits whenever a cluster was successfully deployed to or failed deploying. It will tell you how many clusters are left, how many had issues being deployed to and what these clusters were. This is what the object looks like. Information is purposely made redundant for easier logging where needed.

{
  "percent": 0.5,
  "clusters": {
    "total": 4,
    "completed": 2,
    "found": [
      "cluster-1",
      "cluster-2",
      "cluster-3",
      "cluster-4"
    ],
    "remaining": [
      "cluster-2",
      "cluster-3"
    ],
    "successful": [
      "cluster-1",
      "cluster-2"
    ],
    "failed": []
  }
}

Namespaces

To create a Namespace within a given cluster, simply provide the namespace files within a directory matching the name of the cluster. So if you had a cluster called my-cluster, you could place the following contents into a file called ./namespaces/my-cluster/example-namespace.yaml.

apiVersion: v1
kind: Namespace
metadata:
  name: example

Manifests

All manifest files should be placed into directories corresponding to the name of the cluster + namespace they are for. So for example, if you wanted to deploy manifests to a cluster called my-cluster, you could place all the manifest files into ./manifests/my-cluster/**/*.yaml.

Reserved Labels

The following labels are dynamically generated and managed by kit for all manifests. They should not be manually set and if they are an error will be thrown halting the deploy.

• id
• strategy

Required Labels

The following labels must be set on Deployment manifests and on the labels for the pod template within the Deployment manifest.

• name

Supported Types

Currently we only properly support the following types to be used as manifests. Any other type used will not be deployed and will display a warning:

• Deployment
• Job
• Secret
• Service
• DaemonSet
• PersistentVolumeClaim

Order of Deploys (beta)

NOTE: this feature is a work in progress and may not function correctly

By default, all manifests are deployed to a given cluster at the same time. If however you require manifests to be deployed in a specific order you can utilize the "dependency-selector". By specifying a manifest with dependencies on other resources, the deployer will only deploy that manifest once those other services are fully available on the cluster.

You can specify dependencies using a metadata annotation called kit-deployer/dependency-selector. It should be a valid label selector. For example:

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: proxy-deployment
  annotations:
    kit-deployer/dependency-selector: database=mongo,env=prod

Supported Dependency Types

Currently we only properly support the following types to be used as dependencies:

• Deployment
• Job
• Secret
• Service

Manifest backup

With BACKUP_ENABLED set to true each manifest will be backed up to the specified S3 bucket with the path /AWS_BUCKET/cluster-name/manifest after the manifest has been deployed. By default files can be backed up as yaml or json format - yaml is the default.

Expected environment variables

The following environment variables are used by this service.

Variable	Description	Required	Default
API_VERSION	The kubernetes api version to use	yes	v1
CI_COMMIT_ID	The commit sha that we are deploying. Needed if GITHUB_ENABLED=true	no	empty
SELECTOR	Selector (label query) to filter on	no	empty
CONFIGS	Set the pattern to search for cluster config files	yes	/configs/**/kubeconfig
NAMESPACES_DIR	Set the directory where all the namespace files are. They should be grouped in folders matching the metadata.name of the cluster that you want them deployed to	yes	/namespaces
MANIFESTS_DIR	Set the directory where all the manifest files are. They should be grouped in folders matching the metadata.name of the cluster that you want them deployed to	yes	/manifests
DRY_RUN	Will only show the diff and will not push anything to the cluster	yes	true
UUID	A UUID to be used in webhooks and API requests	false	empty
DEPLOY_ID	A unique id to be used in deployments	false	empty
STRATEGY	The deployment strategy to use. Currently supports rolling-update, fast-rollback.	false	rolling-update
RESOURCE	The resource name for the manifests being deployed	false	empty
DEBUG	A boolean flag to enable debug mode	yes	false
IS_ROLLBACK	A boolean flag that is passed in the available payload post	yes	false
DIFF	Will show a diff	yes	false
FORCE	Will push all changes even if there are no differences	yes	false
CREATE_ONLY	Will only create resources that don't already exist. Will not perform any updates or removals.	yes	false
RAW	Will not perform any alterations to the manifest and simply deploy them as is	yes	false
BACKOFF_FAIL_AFTER	Maximum number of backoffs before the call is aborted	yes	10
BACKOFF_INITIAL_DELAY	The initial delay when starting to backoff in ms	yes	1000
BACKOFF_MAX_DELAY	The maximum delay when doing a backoff in ms	yes	30000
AVAILABLE_ENABLED	Will check if deployed service is available, but will only affect if deployment is considered successful or not if --available-required is also enabled	yes	false
AVAILABLE_POLLING_INTERVAL	The number of seconds to wait between status checking API requests	yes	false
AVAILABLE_HEALTH_CHECK	Will monitor for issues during the deployment	yes	true
AVAILABLE_HEALTH_CHECK_GRACE_PERIOD	The amount of time in seconds that the health check will wait for a deployment to self-heal before triggering a failure	yes	10
AVAILABLE_HEALTH_CHECK_THRESHOLD	The number times a particular error event can occur before triggering a failure	yes	0
AVAILABLE_HEALTH_CHECK_IGNORED_ERRORS	Errors that the built in healthchecker will ignore	yes	empty
AVAILABLE_WEBHOOK	The URL you want to send the status payload of the deployment progress to. You can provide multiple endpoints by using a JSON array of URLs	no	empty
AVAILABLE_ALL	When enabled it will check if all manifests are available even if no differences were deployed	yes	false
AVAILABLE_REQUIRED	Will only finish once the manifest is considered available in the cluster	yes	false
AVAILABLE_KEEP_ALIVE	Will print the status of the available check every AVAILABLE_KEEP_ALIVE_INTERVAL seconds (useful for CI tools that require log output to prevent timeouts)	yes	false
AVAILABLE_KEEP_ALIVE_INTERVAL	Determines the interval at which the keep alive message will be printed	yes	30
AVAILABLE_TIMEOUT	The number of seconds to wait for a given manifest to be available	yes	600
DEPENDENCY_WAIT	The number of seconds to wait between status check attempts for a dependency	yes	3
DEPENDENCY_TIMEOUT	The number of seconds to wait before timing out waiting for a dependency to be available	yes	600
GITHUB_ENABLED	If true, will check the date of the commit against github and will only deploy if the commit is newer than what is on the cluster.	yes	true
GITHUB_AUTH_TOKEN	Your github token to the repo we are deploying (used to retrieve additional info on the commit)	yes	empty
GITHUB_USER	The github user that the repo belongs to	yes	empty
GITHUB_REPO	The github repo name	yes	empty
BACKUP_ENABLED	Enable Backups to AWS S3	no	false
SAVE_FORMAT	Format to save backup manifest as (json	yaml)	no
AWS_BUCKET	AWS s3 bucket name to save to	yes	empty
AWS_ACCESS_KEY_ID	AWS User key to use	yes	empty
AWS_SECRET_ACCESS_KEY	AWS User Secret to use	yes	empty

Contributing

See the Contributing guide for steps on how to contribute to this project.