Revel provides the Jobs framework for performing work asynchronously, outside of the
request flow. This may take the form of recurring tasks that updates cached data
or one-off tasks such as sending emails.
The Jobs framework is included as an optional module, and is not enabled by default.
To activate it, add module.jobs to the app.conf file:
Additionally, in order to access the job monitoring page, you will need to add
this line to the conf/routes file, which will insert the /@jobs url:
There are a couple of configuration settings that tell the framework what sort of limitations
to place on the jobs that it runs. These are listed below with their default values;
To create a Job, implement the cron.Job interface. The
Job interface has the following signature:
To run a task on application startup, use
revel.OnAppStart() to register a function.
Revel runs these tasks serially, before starting the server. Note that this
functionality does not actually use the jobs module, but it can be used to
submit a job for execution that doesn’t block server startup.
Jobs may be scheduled to run on any schedule. There are two options for expressing the schedule:
A cron specification
A fixed interval
Revel uses the cron library to parse the
schedule and run the jobs. The library’s
README provides a detailed
description of the format accepted.
Jobs are generally registered using the
revel.OnAppStart() hook, but they may be
registered at any later time as well.
Here are some examples:
You can configure schedules in the app.conf file and reference them anywhere.
This provides an easy way to reuse, and a useful description for crontab specs.
Here is an example named cron schedule, in an app.conf file:
cron.workhours_15m = 0 */15 9-17 ? * MON-FRI
Use the named schedule by referencing it anywhere you would have used a cron spec.
IMPORTANT: The cron schedule's name must begin with cron.
Sometimes it is necessary to do something in response to a user action. In these
cases, the jobs module allows you to submit a job to be run a single time.
The only control offered is how long to wait until the job should be run.
It is possible to register a func() as a job by wrapping it in the jobs.Func
type. For example:
The jobs module provides a status page (/@jobs url) that shows:
a list of the scheduled jobs it knows about
the current status; IDLE or RUNNING
the previous and next run times
For security purposes, the status page is restricted to requests that originate
Constrained pool size
It is possible to configure the job module to limit the number of jobs that are
allowed to run at the same time. This allows the developer to restrict the
resources that could be potentially in use by asynchronous jobs – typically
interactive responsiveness is valued above asynchronous processing. When a pool
is full of running jobs, new jobs block to wait for running jobs to complete.
Implementation Note: The implementation blocks on a channel receive, which is
implemented to be FIFO for waiting goroutines (but not specified/required to be
so). See here for discussion.
Future areas for development
Allow access to the job status page with HTTP Basic Authentication credentials
Allow administrators to run scheduled jobs interactively from the status page
Provide more visibility into the job runner, e.g. the pool size, the job queue length, etc.