Skip to content
Kapten
Esc
↑↓navigate↵open⌘Jpreview
On this page

Components

Components are what Kapten runs in an environment, applications built from GitHub or pulled from an image, jobs that run once or on a schedule, and databases.

A component is something Kapten runs in an environment. There are three kinds: applications, jobs and databases. Every deployment of a component is a run, with its status, a timeline of its steps and its logs.

Applications

An application is a long-running service, such as a web app, an API or a queue worker. Its source is one of:

  • GitHub: a repository and branch from a connected GitHub installation. Kapten builds a container image from the repository with Railpack, which detects the language and framework, and can rebuild automatically on every push to the branch.
  • Container: an image name and tag from a container registry.

An application also defines:

Setting Description
Resources CPU and memory requests and limits, and an optional GPU count. See Resources.
Instances Minimum and maximum number of running instances.
Ports Internal ports the application listens on, each with a protocol: HTTP, GRPC, TCP or UDP. One port is the default.
Storage Persistent volumes, each with a mount path, a size in GiB and a Kubernetes storage class.
Health checks A readiness probe, which decides when an instance receives traffic, and a liveness probe, which restarts an instance that stopped responding. Each is an HTTP, TCP, gRPC or command check.
Runtime command An entrypoint and arguments that replace the image’s own.

Jobs

A job runs a container until it exits. It is either:

  • cron: run on a schedule written in the standard five-field cron syntax (minute hour day month weekday), in UTC unless you set a time zone;
  • one-time: run once, when you deploy it.

A job sets the same resources as an application, plus the maximum number of restarts allowed before the run fails (0 allows none) and the maximum duration after which it is stopped and marked failed.

Databases

A database is a PostgreSQL, MySQL, MongoDB or Redis instance. It is provisioned either:

  • in cluster: Kapten runs the database on the environment’s cluster;
  • managed: Kapten provisions the database service of your cloud provider.

You choose its engine version, whether it is reachable only from inside the cluster (PRIVATE) or from the internet (PUBLIC), its resources, and whether to enable disk encryption, high availability and backups.

Kapten generates the database’s credentials and adds them to the environment as secrets, so your applications can connect without copying passwords around.

Was this page helpful?