kuda

kuda

Simple deployment tool for monorepo

Commands

init

This is run at the root of the repository to initialize kuda. You will be asked to select a service (folder) you wish to have managed under kuda.

kuda init

add

Once a repository is initialized, a kuda.json will be generated in the root of the repository. You can continue adding services using the add command or alternatively, update kuda.json or package.json to reflect the changes you want.

kuda add

run

This command will attempt to run all the tasks for each service defined in kuda.json. Once all tasks have completed successfully, the remoteState is updated.

The localVersion (derived from package.json) and the remoteVersion are compared before deployment. If the remoteVersion is up to date, no action will be taken.

Thus, in order to trigger tasks to run, the user is required to increment the version inside <service>/package.json

KUDA_ENV=staging kuda run

You can now target a specific service using

KUDA_ENV=staging kuda run -s foo

Typical folder structure

monorepo
├── foo
│   └── package.json
├── bar
│   └── package.json
├── unmanagedFolder1
├── unmanagedFolder2
├── kuda.json

kuda.json

  services: [
    {
      name: 'foo'
    },
    {
      name: 'bar'
    }
  ],
  remoteState: s3://monorepo/{KUDA_ENV}/kudastate

This will run task(s) for foo and then task(s) for bar.

foo/package.json

name: 'foo'
version: 0.0.1,
kuda: {
  tasks: [
    {
      name: 'foo',
      command: 'echo {SOME_ENV}'
    },
    {
      name: 'bar',
      command: 'cat file.json stuff.json'
    }
  ]
}

remoteState

services: {
  foo: {
    deployedVersion: '0.0.0'
  },
  bar: {
    deployedVersion: null
  }
}

TODO

child process maxBuffer issue. might need to stream-- Increased buffer to fix

• Add some tests...

• currently change detection is based on package.json,

can swap out for other means like using git commits (compare with s3 stored remote), git tags, timestamps

• allow service and version name to be accessible in tasks (attach to process.env after reading values)

(ie <SERVICE_NAME> <SERVICE_VERSION>)

• allow users to control deployment flow by creating dependency tree

• allow different types of task (or not) (ie. tasks that are run regardless of whether there has been change)