How to use the Faktory plugin

This plugin lets Kestra flows submit background jobs to a Faktory server, the language-agnostic job server most commonly paired with Sidekiq / Ruby on Rails workers. It talks to Faktory directly over the Faktory Work Protocol (FWP) - a line-based TCP protocol, not HTTP - so no third-party client library is bundled.

Authentication

Every task connects using:

• host - hostname or IP address of the Faktory server, defaults to localhost.
• port - TCP port the server listens on, defaults to 7419.
• password - required only if the server is password-protected. Store it as a secret and reference it with {{ secret('FAKTORY_PASSWORD') }} - never hardcode it in a flow.
• tls - set to true to connect over TLS (tcp+tls://), defaults to false. The connection verifies the server's certificate against the JVM's trust store, so a server presenting a self-signed certificate will fail to connect unless that certificate is added to the trust store.

On connect, the server greets the client with its protocol version and, if password-protected, a salt and iteration count. The plugin replies with the SHA256-iterated password hash automatically; the plain password itself is never sent over the wire or logged.

Connection properties are shared across tasks, so consider setting them once via plugin defaults if you have several Faktory tasks in the same namespace.

Tasks

Push

Enqueues (pushes) a single job to a Faktory queue and returns its jid (job id) as output, so downstream tasks can reference it. It is fire-and-forget: the task closes the connection right after the server acknowledges the job, it does not wait for a worker to process it.

Required:

• jobType - the Faktory jobtype identifying which worker handles the job.

Optional:

• args - JSON-native arguments passed to the worker, defaults to an empty list.
• queue - the queue to push to, defaults to default.
• retries - how many times Faktory retries the job before moving it to the dead queue (-1 to 1000), defaults to 25. A negative value discards the job on failure instead of retrying it or moving it to the dead queue.
• reserveFor - how long a worker may hold the job before Faktory reschedules it, minimum PT60S, defaults to PT30M.
• at - schedules the job for a future instant instead of making it immediately available.
• jobId - a custom jid; if omitted, a random UUID is generated and returned as the output.
• custom - arbitrary metadata attached to the job payload, available to the worker at fetch time.

id: faktory_push
namespace: company.team

inputs: 
  - id: email
    type: STRING

tasks: 
  - id: enqueue_welcome_email
    type: io.kestra.plugin.faktory.Push
    host: "{{ secret('FAKTORY_HOST') }}"
    password: "{{ secret('FAKTORY_PASSWORD') }}"
    jobType: SendWelcomeEmail
    queue: mailers
    args: 
      - "{{ inputs.email }}"

Triggers

None yet. Consuming jobs from a Faktory queue (FETCH/ACK/FAIL) is tracked as a separate, future addition.