Infinitas

A scheduler with persistence. Useable through both a node.js and a HTTP API.

Note: Infinitas is in active development. This Readme currently describes what version 1.0.0 of Infinitas will provide.

What is working so far ?

• In memory scheduling with 'cron' and 'interval' functionality.
• No PostgreSQL support
• Infinitas runs with only a single master node. It is not distributed and resilient yet

Overview

Infinitas is a node.js based scheduling engine. Think of it as an application level 'cron' scheduler. It exposes a node.js client library as well as a websocket API and therefore can be integrated with virtually any application supporting websockets.

What can Infinitas do for you ?

If your application has tasks that need to occur at specific moments in time or if you need regular maintenance jobs to be triggered, Infinitas is for you.

A few examples of what Infinitas can do:

• Perform database archival at regular interval (eg. at midnight or every 12 hours)
• Crawl your database once a week for records that can be archived and publish these in an other, less dynamic storage
• Provide calendar reminders for your application (eg. send an email 1 day before a task is due)

Infinitas features include:

• persistence: it makes your jobs resilient (out of the box with postgres)
• monitoring/audit: it comes with a user interface for your jobs statuses and can notify you of failures
• scalable: run several instances to handle real production load

The basics

There are 2 ways that you can deploy Infinitas. It can be embedded within your node.js instances or it can be deployed as a standalone server with multiple clients.

Quick start

Some environment variables can be set to point infinitas to the proper PostgreSQL database:

• INFINITAS_DB_HOST defaults to 127.0.0.1
• INFINITAS_DB_PORT defaults to 5432
• INFINITAS_DB_NAME defaults to infinitas
• INFINITAS_DB_USER defaults to infinitas
• INFINITAS_DB_PWD defaults to infinitas

You can also set the port used by Infinitas through INFINITAS_PORT.

const Infinitas = require('infinitas')
    
var infinitas = new Infinitas()

var task = {
  name: 'myTaskName',
  schedule: '*/15 * * * *',
  timeout: 60000 // timeout in milliseconds
}

infinitas.schedule(task, function(err) {
  // jobs for that task will be created every minute and every hour
})

// Here is how you declare your business logic
infinitas.addProcessor(task.name, function(job) {
  // task business logic here...

  job.on('timeout', (taskTimeoutValue) => {
    // The job timed-out
    // Note that because this event, job.done(), and job.fail() are all
    //   asynchronous there is no guarantee that the timeout event will fire
    //   before you call job.done() or job.fail().
    // Regardless of a job timeout status, job.log() will continue to log
    //   your messages.
  })
  
  // log information related to the job
  // An optional callback will catch transmission errors to the server.
  // If no callback is used, errors are ignored
  job.log('A Fox once saw a Crow...')

  // when the task is finished, call this.done()
  job.done((err) => {
    // err is non null when:
    //   - infinitas server cannot be contacted (err.code = 'server_unreachable')
    //   - the job has already timed out (err.code = 'job_timed_out')
    //   - job.done() was already called (err.code = 'job_done')
    //   - job.fail() was previously called (err.code = 'job_failed')
  })
      
  // or if it failed, you can call this.fail()
  job.fail((err) => {
    // err is non null when:
    //   - infinitas server cannot be contacted (err.code = 'server_unreachable')
    //   - the job has already timed out (err.code = 'job_timed_out')
    //   - job.fail() was already called (err.code = 'job_failed')
    //   - job.done() was previously called (err.code = 'job_done')
  })


  // Note that you cannot log messages after calling either job.done() or
  //   job.fail()
  job.log('A Fox once saw a Crow...') // this will throw an error

})

Objects

As an Infinitas consumer, you will be defining Tasks and Processors. At a specific time or interval, a Task is sent as a Job to a Processor.

Tasks

A task has the following properties:

| Property | -sub | Type | Description | |---------------|--------|---------|--------------------------------------------------------------------------| | name | | string | each task is uniquely identifiable through a name within the system | | lastJob * | | Object | an object representing the last triggered job | | | id | string | the unique identifier of the last triggered job | | | date | Date | represents the date of the last triggered job | | | status | string | one of "succeeded", "failed", "running" | | lastFailure * | | Object | an object representing the last failed job. Attributes same as above | | lastSuccess * | | Object | an object representing the last successful job. Attributes same as above | | schedule | | string | CRON-like schedule |

Properties with a * are read only.

Job

Processor

Multiple clients

NOT SUPPORTED YET !

The above quick start runs a single client and a single Infinitas server in the local process. If you want multiple clients to concurrently run tasks, Infinitas can be used by multiple clients.

In that configuration, each client opens a websocket connection to the server. Job allocation to clients is round-robin.

On each client:

  const Infinitas = require('infinitas')
    
  var infinitas = new Infinitas({
    server: 'wss://my-infinitas-server:3000'
  })

  infinitas.processor('myTaskName', function(job) {
    // do smthg
    job.done()
  })

If no processor is active, the server will retry after 1, 10 and 30 seconds.

To start the server, download infinitas and run npm start. Our goal is to provide a deb package but that work is still in progress.

Database

Infinitas creates and maintain its own tables. Migrations are run when the server is started. All tables are prefixed with infinitas_ to prevent conflicting with other database users.