1. Mahlon Smith
  2. symphony-metronome




Metronome is an interval scheduler and task runner. It can be used locally as a cron replacement, or as a network-wide job executor.

Events are stored via simple database rows, and optionally managed via AMQP events. Interval/time values are expressed with intuitive English phrases, ie.: 'at 2pm', or 'Starting in 20 minutes, run every 10 seconds and then finish in 2 days', or 'execute 12 times during the next minute'.


Here's an example of a simple cron clone:

require 'symphony/metronome'


Symphony::Metronome.run do |opts, id|
    Thread.new do
        pid = fork { exec opts.delete('command') }
        Process.waitpid( pid )

And here's a simplistic AMQP message broadcaster, using existing Symphony connection information:

require 'symphony/metronome'


Symphony::Metronome.run do |opts, id|
    key = opts.delete( 'routing_key' ) or next
    exchange = Symphony::Queue.amqp_exchange
    exchange.publish( 'hi from Metronome!', routing_key: key )

Adding Actions

There are two primary components to Metronome -- getting actions into its database, and performing some task with those actions when the time is appropriate.

By default, Metronome will start up an AMQP listener, attached to your Symphony exchange, and wait for new scheduling messages. There are two events it will take action on:


Create a new scheduled event.  The payload should be a hash.  An
'expression' key is required, that provides the interval description.
Anything additional is serialized to 'options', that are passed to the
block when the interval fires.  You can populate it with anything
your task requires to execute.


The payload is the row ID of the action.  Metronome removes it from
the database.

If you'd prefer not to use the AMQP listener, you can put actions into Metronome using any database methodology you please. When the daemon starts up or receives a HUP signal, it will re-read and schedule out upcoming work.


Metronome uses Configurability to determine behavior. The configuration is a YAML file. It shares AMQP configuration with Symphony, and adds metronome specific controls in the 'metronome' key.

    splay: 0
    listen: true
    db: sqlite:///tmp/metronome.db


Randomize all start times for actions by this many seconds on either side of the original execution time. Defaults to none.


Start up an AMQP listener using Symphony configuration, for remote administration of schedule events. Defaults to true.


A Sequel connection URI. Currently, Metronome is tested under SQLite and PostgreSQL. Defaults to a SQLite file at /tmp/metronome.db.

Scheduling Examples

Note that Metronome is designed as an interval scheduler, not a calendaring app. It doesn't have any concepts around phrases like "next tuesday", or "the 3rd sunday after christmas". If that's what you're after, check out the chronic library instead.

Here are a small set of example expressions. Feel free to use the metronome-exp utility to get a feel for what Metronome anticipates.

in 30.5 minutes
once an hour
every 15 minutes for 2 days
at 2014-05-01
at 2014-04-01 14:00:25
at 2pm
starting at 2pm once a day
start in 1 hour from now run every 5 seconds end at 11:15pm
every other hour
run every 7th minute for a day
once a day ending in 1 week
run once a minute for an hour starting in 6 days
10 times a minute for 2 days
run 45 times every hour
30 times per day
start at 2010-01-02 run 12 times and end on 2010-01-03
starting in an hour from now run 6 times a minute for 2 hours
beginning a day from now, run 30 times per minute and finish in 2 weeks
execute 12 times during the next 2 minutes
once a minute beginning in 5 minutes

In general, you can use reasonably intuitive phrasings. Capitalization, whitespace, and punctuation doesn't matter. When describing numbers, use digit/integer form instead of words, ie: '1 hour' instead of 'one hour'.


gem install symphony-metronome


You can check out the current development source with Mercurial via its project page.

After checking out the source, run:

$ rake

This task will run the tests/specs and generate API documentation.

If you use rvm, entering the project directory will install any required development dependencies.