Deployment

helm upgrade --install \
    --namespace=your-namespace \
    your-app -f values.yml larakube/larakube

The following command can be used to force a redeployment of the application.

kubectl rollout restart deployment --namespace your-namespace your-app-web your-app-worker-default

This can be useful for example in one of the following cases:

• You changed the value of a secret and want to apply it to the application
• The image of the application was updated, but the tag of the image did not change (f.e. the tag is latest or develop).

Queue workers

Every entry under worker creates a Deployment named <release>-worker-<name>. By default the worker container runs the image’s entrypoint, like the web container. If your image does not pick its role from an environment variable, set the command of the worker:

worker:
  default:
    command: ["php", "artisan", "queue:work"]
    args: ["--tries=3", "--max-time=3600"]
  high:
    replicaCount: 2
    command: ["php", "artisan", "queue:work"]
    args: ["--queue=high"]

Run commands after a deployment

LaraKube can run commands once after every helm install and helm upgrade, for example to clear the application cache. The commands run in a separate Job that uses the same image, environment and secrets as the application.

postDeployment:
  enabled: true

By default the Job runs php artisan cache:clear. You can replace the command:

postDeployment:
  enabled: true
  command:
    - /bin/sh
    - -c
    - php artisan cache:clear && php artisan queue:restart

If database migrations are enabled, the post-deployment Job starts after the migration Job finished successfully. If the command fails, helm upgrade reports the deployment as failed. Use postDeployment.retry to retry it.

[!NOTE] The Job runs in its own pod. Commands that only write files inside the container, like config:cache, route:cache or optimize, do not affect the web and worker pods. Run those when building the image instead. Use the post-deployment Job for shared state like the cache store, the database or the queue.

Helm runs the Job once the new resources are applied, which can be before every new pod is ready. Add --wait to helm upgrade if the command must run after the rollout has finished.

Database migrations

Enable databaseMigration to run php artisan migrate --isolated --force in a Job after every install and upgrade:

databaseMigration:
  enabled: true