Skip to main content

Deploy jobs

A deploy job is a command that is executed before a new release gets deployed. The rollout of the release will only continue if the deploy job finished successfully. The command of the deploy job has access to the same environment and binaries as the app runtime. A common use-case for a deploy job is to run database schema migrations. It can either be specified on creation of the app or later using the update command.

If a deploy job fails, the associated release will be set to failed and the previous release will continue to run if there was one to begin with.

Specifying a deploy job on app creation​

You can specify a deploy job when creating an application by using various deploy job related flags on the nctl create app command.

--deploy-job-command="rake db:prepare"       Command to execute before a new release gets deployed. No deploy job will be executed
                                             if this is not specified.
--deploy-job-name="release"                  Name of the deploy job. The deployment will only continue if the job finished successfully.
--deploy-job-retries=3                       How many times the job will be restarted on failure. Default is 3 and maximum 5.
--deploy-job-timeout=5m                      Timeout of the job. Default is 5m, minimum is 1 minute and maximum is 30 minutes.

$ nctl create app --deploy-job-command="rake db:prepare"

The deploy job will run whenever the new release requires the app to be restarted. The following events will trigger the deploy job to run:

  • A new build is available
  • The configuration of the application changed (new environment variables, size change, etc)

Specifiying a deploy job on other configuration layers​

As deploy jobs are part of the configuration of a deplo.io application, you can also specify them on other configuration layers. For example, you can specify a deploy job via the git configuration source file .deploio.yaml:

deployJob:
  name: "release"
  command: "rake db:prepare"
  retries: 3
  timeout: 5m

Viewing the status of a deploy job​

If a deploy job fails, the associated release will be set to failed and the previous release will continue to run if there was one to begin with. To see the detailed status of a deploy job you can get the full release:

$ nctl get releases brave-scrambler-vc78b -o yaml
[...]
status:
  atProvider:
    deployJobStatus:
      exitTime: 2023-07-18T11:01:47Z
      name: brave-scrambler-vc78b-deploy-job
      reason: backoffLimitExceeded
      startTime: 2023-07-18T11:00:58Z
      status: failed
    releaseStatus: failure

At the bottom of the release you can see the status and it will show in detail when and how a deploy job failed. In addition to the status, the deploy job's log will be written to the normal app log and can be accessed using the nctl logs app command.