Pipeline

The pipeline is the ordered sequence of tasks executed on each server during a deployment.

Run cata pipeline to list the current pipeline and inspect the task order.

Default pipeline

Task	Description
deploy:lock	Creates a lock file to prevent concurrent deployments
deploy:release	Creates the release directory
deploy:update_code	Transfers code to the server (provided by recipe or custom task)
deploy:shared	Symlinks shared_dirs and shared_files into the release
deploy:publish	Switches the current symlink to the new release
deploy:log_revision	Records the deployment as JSON in .catapult/revisions.log
deploy:healthcheck	Checks that the application is responding
deploy:unlock	Removes the lock file (also called on failure)
deploy:cleanup	Removes old releases

deploy:healthcheck is automatically removed if no host defines a healthcheck.

deploy:update_code is intentionally empty in the core package. Import a recipe such as git, rsync, or astro, or override the task yourself.

For complete examples with official recipes, see Deployment Examples.

Available tasks (not in pipeline by default)

These tasks are registered but not inserted automatically. By default they run in {{release_path}}.

Task	Default implementation	Description
deploy:install	pmInstall() in {{release_path}}	Installs dependencies
deploy:build	pm() run build in {{release_path}}	Builds the app
deploy:test	pm() run test in {{release_path}}	Runs tests

import { after } from '@catapultjs/deploy'

after('deploy:update_code', 'deploy:install')
after('deploy:install', 'deploy:build')

Override the implementation if needed:

import { task, cd, run } from '@catapultjs/deploy'

task('deploy:build', () => {
  cd('{{release_path}}')
  run('node ace build')
})

Adding a task

Use after() or before() to insert a task relative to an existing one:

import { task, cd, run, after, before } from '@catapultjs/deploy'

task('cache:clear', () => {
  cd('{{current_path}}')
  run('node ace cache:clear')
})

after('deploy:publish', 'cache:clear')
before('deploy:cleanup', 'cache:clear')

Removing a task

import { remove } from '@catapultjs/deploy'

remove('deploy:healthcheck')

Checking if a task is in the pipeline

Use inPipeline() to attach a task relative to optional pipeline steps:

import { inPipeline, after } from '@catapultjs/deploy'

if (inPipeline('deploy:healthcheck')) {
  after('deploy:healthcheck', 'notify')
} else {
  after('deploy:publish', 'notify')
}

Replacing the entire pipeline

import { setPipeline } from '@catapultjs/deploy'

setPipeline([
  'deploy:release',
  'deploy:update_code',
  'deploy:shared',
  'deploy:publish',
  'deploy:unlock',
  'deploy:cleanup',
])

Overriding a task

Redefining a task replaces its implementation:

import { defineConfig, task, cd, run } from '@catapultjs/deploy'

task('deploy:update_code', () => {
  cd('{{release_path}}')
  run('rsync ...')
})

export default defineConfig({ ... })

Async task

For operations that require more than SSH commands, use an async function and destructure the TaskContext parameter:

import { type TaskContext, task, after } from '@catapultjs/deploy'

task('notify', async ({ release }: TaskContext) => {
  await fetch(process.env.SLACK_WEBHOOK!, {
    method: 'POST',
    body: JSON.stringify({ text: `Deployed ${release}` }),
  })
})

after('deploy:healthcheck', 'notify')

Running a task manually

Any registered task can be run directly from the terminal:

npx cata task deploy:update_code
npx cata task cache:clear --host staging

Template variables

Available in cd() and run():

Variable	Value
{{release_path}}	/base/releases/<release>
{{current_path}}	/base/current
{{shared_path}}	/base/shared
{{releases_path}}	/base/releases
{{base_path}}	/base
{{release}}	Release name (e.g. 2024-01-15T10-30-00-000Z)

Where /base is the deployPath defined on the host.

import { task, cd, run } from '@catapultjs/deploy'

task('my:task', () => {
  cd('{{release_path}}')
  run('cp {{shared_path}}/.env {{release_path}}/.env')
  run('echo "Release: {{release}}"')
})