Skip to content

Latest commit

 

History

History
206 lines (161 loc) · 8.41 KB

README.md

File metadata and controls

206 lines (161 loc) · 8.41 KB
Raw

konfigure

This tool facilitates deploying workloads to K8s clusters, providing several convenience features streamlining developer workflows and guarding against common failure modes. It is to be used in conjunction with a Helm/CDK8s chart repository and a live application configuration repository, or within a single application repository.

Installation

This tool can be installed using npm:

npm install -g konfigure

Usage

The konfigure command can be used to render charts deploy them to a cluster, and to teardown afterwords. It is to be run in a directory with a konfig.yaml file which sets up an environment.

Commands

The basic form these commands take is:

konfigure [flags...] [subcommand] [environment] [(optional: chart) target... | all]

with the components being as follows:

  • flags: keywords which, when set, modify the operation of the function
    • auth: assumes the appropriate AWS role
    • debug: enables more detailed logging
    • dryrun simulates a run, printing out commands that would be executed
    • testing disables validations and updates, to allow faster iteration
    • cd indicates that we are in a Continuous Deployment (CD) environment, enables support for cdDisabled
  • subcommand: see commands
  • environment: which environment to operate in, corresponds to a directory
  • target: which deployments to operate on. if all, no limiting. if prepended with chart, args will be interpreted as chart names, rather than deployment names.

Command Topics

Examples

This command pushes the eth-node deployment to the dev environment:

konfigure deploy dev eth-node

This command does almost the same thing, but prints out the actions that would have been done instead:

konfigure --dryrun deploy dev eth-node

This command deploys all instances of the backend chart in the prod environment, ie, if releasing a new version:

konfigure deploy prod chart backend

This command generates the manifests for all deployments in the dev environment and outputs them to a single file:

konfigure render dev all > dev-release.yaml

This command uninstalls the seeder and backend deployments from the dev environment:

konfigure teardown dev seeder backend

Configuration

Environments are configured within a directory called env in the root of a repository; each environment is represented by a subdirectory and contains a configuration file, as well as several subdirectories, each of which maps to a field in the configuration file.

Configuration File

The konfig.yaml file sets up the environment, including the relevant account, region, cluster, namespace, and node group. It also configures each deployment, setting the name, chart, and source.

For backwards-compatibility reasons konfigure will recognize any file matching {k,c}onfig.{yaml,json}, but only konfig.yaml will be used moving forward.

Fields
  • apiVersion: minimum version of chitin required
  • environment: environment configuration
    • awsAccount (required): AWS account to use
    • awsRegion (optional): AWS region to use. if not set, detection will be attempted
    • k8sContext (required): pre-configured K8s context to use
    • k8sNamespace (required): K8s namespace to create/use
    • eksNodegroup: EKS nodegroup to deploy to
    • tfEnv: name of the Terraform environment
    • tfModule: Terraform module to reference
  • chartDefaults: defaults to apply to any instance of a chart, fields same as deployments applies values to all deployments of a certain chart in an environment, ie versions, resources, etc
  • deployments (required): workloads to be deployed, each is an instance of a chart with these fields:
    • chart: name/path of the chart to use
    • type: the type of chart (Helm/CDK8s) [default: Helm]
    • source: where to pull the chart from, options are:
      • local: local filesystem, expects a path (relative or absolute)
      • artifactory: pull from fimbulvetr, expects the repo to be configured with Helm
      • remote: pull from a remote repository
    • version: version of the chart to require [optional]
    • values: an inline values object, will be converted to YAML and passed to the chart [optional]
    • disabled: if set to true, causes this deployment to not be processed [default: false]
    • cdDisabled: if set to true, and the cd flag is set, causes this deployment to not be processed [default: false]
  • externalResources: enables creation of K8s Services/Secrets to facilitate access to extra-cluster resources, ie databases, MSK clusters, Hetzner services, or external APIs
    • secretPresets: defines common externalSecret combinations which can be referenced in specific externalResources.deployments using $secretPreset
    • deployments: instances of the external-service chart to be deployed, takes only values
Secrets Management

We use the external-secrets module to poplate K8s Secrets from parameters stored in AWS SSM, which allows us to avoid manually shuffling around sensitive values or accidentally committing them to Git. To use this functionality, we create an ExternalSecret resource with a map of secret key to SSM parameter path, and then reference the automatically created Secret using a secretEnvMap block or the equivalent.

Note that the external-service chart provides an easy way of setting up these resources by adding an item to the externalResources.deployments section, see the example below for details.

Subdirectory Mappings
  • chartDefaults: files map to the equivalent config entry's values field
  • deployments: files map to the equivalent config entry's values field
  • externalResources: files map to the equivalent config entry's value (in the deployments subsection)
Overrides & Precedence

The subdirectories/files are simply a way to extract complex configuration from the main file, and are merged with inline config in this precedence order (higher overrides):

  1. chart defaults (values.yaml)
  2. chartDefaults (file)
  3. chartDefaults (inline)
  4. deployments (file)
  5. deployments (inline)
Example
apiVersion: 4.13.0
environment:
  tfEnv: dev
  tfModule: legacy-dev
  awsAccount: dev
  awsRegion: eu-east-1
  k8sContext: nonprod
  k8sNamespace: dev
  eksNodegroup: eu-east-1a-workers

chartDefaults:
  external-service:
    version: 0.1.0
  ethereum:
    version: 1.3.3
  backend:
    version: 0.5.4
    values:
      cluster:
        version: 6.47.0

deployments:
  eth-node:
    chart: ethereum
  eth2-node:
    chart: ethereum
  eth-backend:
    chart: backend
  eth-seeder:
    chart: ~/Developer/Projects/charts/charts/seeder
    source: local
  eth-producer:
    chart: producer
    version: 0.5.4
  some-other-app-instance:
    chart: some-other-app
    type: cdk8s
    version: 1.2.3

externalResources:
  secretPresets:
    preset-1:
      username: /dev/some-param/username
      password: /dev/some-param/password
  deployments:
    alchemy-eth:
      externalName: '[...].alchemy.com'
      externalSecrets:
        username: /dev/alchemy-eth/username
        password: /dev/alchemy-eth/password
    postgres-local:
      externalIP: 10.0.0.18
      externalPort: 5432
    postgres-eth:
      externalName: dev-eth.[...].rds.amazonaws.com
      $secretPreset: preset-1
    producer-kafka-1:
      externalName: b-1.kafka-nonprod-producer.[...].amazonaws.com