Porker NPM | npm.io

Porker

A super simple job queue for Node.js (7.6 and higher, uses async functions) built on top of Postgres (9.6 and higher since we use SELECT FOR UPDATE SKIP LOCKED and ADD COLUMN IF NOT EXISTS)

Usage

`new Porker({ connection, queue, errorThreshold, retryDelay, timeout, concurrency, healthcheckPort })`

Creates and returns a new worker.

connection: (optional) Postgres connection details, passed through to new Client in pg
queue: required Name of the job queue, will be represented as a table with a name in the form of ${queue}_jobs
errorThreshold: (optional) Maximum error_count a job can have to be received by the retry handler if one is specified, defaults to 1
retryDelay: (optional) Interval to wait between failure and retry attempt, specified as a Postgres interval string, defaults to '5 minutes'
timeout: (optional) Maximum time a worker function can run before a job will be flagged as an error and returned to the queue, defaults to 15000 (15 seconds)
concurrency: (optional) Maximum number of jobs to process simultaneously, defaults to 1.
healthcheckPort: (optional) If defined, will start a simple healthcheck listener on this port. Returns an HTTP 200 when the worker is connected to postgres, and a 400 when not. Defaults to null

`porker.connect()`

AsyncFunction that connects to the database. This is unnecessary to call manually as all methods that require a connection will call it themselves.

`porker.create()`

AsyncFunction that creates or updates the table representing the current queue. It is recommended to run this method at least once whenever updating Porker as it may create new columns in your job tables.

The SQL run is as follows (replacing ${queue} with the name of your queue):

CREATE TABLE IF NOT EXISTS "${queue}_jobs" (
  id serial PRIMARY KEY,
  priority integer NOT NULL DEFAULT 0,
  started_at timestamp with time zone,
  repeat_every interval,
  error_count integer NOT NULL DEFAULT 0,
  args jsonb
);

ALTER TABLE "${queue}_jobs" ADD COLUMN IF NOT EXISTS retry_at timestamp with time zone;

CREATE INDEX IF NOT EXISTS "${queue}_jobs_error_count_index" ON "${queue}_jobs" (error_count);
CREATE INDEX IF NOT EXISTS "${queue}_jobs_priority_index" ON "${queue}_jobs" (priority);
CREATE INDEX IF NOT EXISTS "${queue}_jobs_repeat_every_index" ON "${queue}_jobs" (repeat_every);
CREATE INDEX IF NOT EXISTS "${queue}_jobs_retry_at_index" ON "${queue}_jobs" (retry_at);
CREATE INDEX IF NOT EXISTS "${queue}_jobs_started_at_index" ON "${queue}_jobs" (started_at);

`porker.drop()`

AsyncFunction that drops the table representing the current queue. THIS WILL DESTROY DATA.

`porker.subscribe(fn)`

AsyncFunction that adds a subscription to the current queue.

The parameter, fn should be an AsyncFunction that accepts a single parameter which will be an object representing the full row from the database for the current job. If the function throws, the job's error_count column will be incremented and it will be returned to the queue. If the function runs successfully, the row will be deleted.

`porker.retry(fn)`

AsyncFunction that adds a retry handler to the current queue.

The parameter, fn should be an AsyncFunction that accepts a single parameter which will be an object representing the full row from the database for the current job. Jobs passed to this worker will have failed at least once, but not more than errorThreshold times. A job that succeeds this handler will be cleared from the database as though it succeeded the first time. If a recurring job is delivered to the retry handler and succeeds there, the error_count will be reset to 0.

`porker.publish(jobs, { priority, repeat })`

AsyncFunction that publishes a new job or jobs to the queue and returns an array of ids. The first parameter may be a single object to publish one job or an array of objects to publish multiple jobs, however each job must be serializable as a jsonb column. The second parameter is entirely optional and is used for passing additional options.

priority: (optional) A priority for the current job, higher priorities will be processed sooner. Defaults to 0
repeat: (optional) For creating a recurring job, can be specified as any Postgres interval See section 8.5.4. Omitting this option creates a standard non-recurring job.

`porker.unpublish(ids)`

AsyncFunction that unpublishes a job or jobs. May pass either a single job id or an array of job ids. This is mostly useful for canceling recurring jobs.