Jobber

Do you like Lambda, the concept of disposable functions? Well this is surely the solution for you! Do you find OpenFaaS overly complicated for a basic setup? Me too! Are you a fan of basic docker-compose setup scripts? Ai Ai!

The objective of this project is to provide nothing more than a basic interface to create lambda-like functions. These functions can be triggered by a HTTP gateway, MQTT events, or cron schedules!

---

Terminology

• Jobber: This application/service
• Job: An individual job, containing multiple versions, runners, actions, triggers and environment variables.
• Action: Describes how a job should run, timeouts, parallelism, run-duration, etc.
• Trigger: Describes how a job should start. MQTT, HTTP, or a Schedule.
• Environment: Describes environment variables for a job.
• Runners: Describes a action that is currently running.
• Runner Manager: Describes the component that controls the connections between runners, actions and triggers.
• Gateway: Describes the http server we utilise for forwarding traffic to jobs and runners.

---

Screenshots

Click Here to see some demo pictures of the application in action!

---

Metrics (Prometheus)

Click Here For more detailed explanation of metrics!

---

Environment Variables

Click Here For more extensive list of environment variable configuration!

• DATABASE_URL Postgres connection URL. Example: postgresql://user:pass@host/db
• JOBBER_NAME The name of your jobber instance, should be unique per host.
• MANAGER_PORT Port that runner-manager server operates on. Default: 5211
• MANAGER_HOST Host that runner-manager server operates on. Default: hostname()
• STARTUP_USERNAME The administrator account username. Created at every startup. Has full permissions. If you change this after a previous start, it will create a NEW account, not update the previous account.
• STARTUP_PASSWORD The administrator account password.

---

Docker Compose

See docker-compose.yaml for a sample setup. Networking is configured securely, volumes are created, and all essential details are provided there.

---

HTTP Ports

For security reasons, this application hides behind two separate ports. One is for the administrative panel, and the other is for the HTTP API Gateway for function invocations.

Port 3000: Administrative panel. Port 3001: HTTP Gateway (Router for HTTP triggers)

---

Logging (Loki and Database)

It is highly suggested to use loki as your log driver if you have a decent amount of log-spam, especially high bandwidth logging. This will defer a lot of load from the database.

---

Creating your first function

Documentation for this will be finalised in the future, though don't hold me to that. We will be providing a jobber-template repository which includes everything you need to create your first TypeScript job. In the meantime, feel free to reference the examples we have provided in the /Examples Directory. We have an example for http-typescript and a schedule-javascript application.

---

API

For further details on the API, please refer to the bruno scripts.

# GET /api/job/

Fetches list of jobs

# GET /api/job/:jobId

Fetches a specific job

# GET /api/job/:jobId/environment/

Fetches a job environment

# POST /api/job/:jobId/environment/:name

Creates a new environment variable. Body should be multi-part form with the following properties:

type = secret|text
value = {variable value}

# DELETE /api/job/:jobId/environment/:name

Deletes new environment variable.

# GET /api/job/actions

Fetches job actions

# GET /api/job/actions:latest

Fetches actions for the latest version

# GET /api/job/triggers

Fetches job triggers

# GET /api/job/triggers:latest

Fetches triggers for the latest version

# GET /api/job/logs

Fetches logs associated with a job

# GET /api/job/runners

Fetches all the current runners associated with a job

# GET /api/job/:jobId/actions/:actionId/runners

Fetches all the current runners associated with an actionId

# DELETE /api/job/:jobId

Deletes a job

# POST /api/job/publish

Publishes a archive (zip) file to jobber. Jobber automatically extracts metadata from the package.json file. If a job does not currently exist, one will be created. This is ideal for CI/CD.

The body should be a multi-part form, with the file attached to the name archive.

See Bruno for further information.

---

Security considerations

• Jobber requires access to a docker daemon, be it docker-in-docker, or the hosts docker daemon. We suggest defaulting to the hosts docker daemon, as its the only tested method. Managing the networking of DIND will be very difficult.