@title Scheduling Guide
Scheduling Asynchronous Jobs
TorqueBox provides support for registering jobs to execute asynchronously on a schedule.
Scheduled jobs are provided by the
torquebox-scheduling gem, and can
be used independently of other TorqueBox services.
A job is basically a zero-arity block that will be executed based on a
schedule you provide. To schedule a job, provide an id, a schedule,
and your block to
TorqueBox::Scheduling::Scheduler.schedule(:job1, every: 1000) do puts "I fire every second" end
Jobs can be scheduled with a delayed start, to execute a limited number of times or up until a stop time, or scheduled based on a cron-like schedule. For more details on the available schedule options, see the Scheduler.schedule docs.
Jobs can be scheduled at runtime, from anywhere within an application.
Using ActiveSupport extensions
If you are using ActiveSupport, you can use its numeric extensions with any of the options that take a time period:
TorqueBox::Scheduling::Scheduler.schedule("my-job", in: 5.minutes, every: 1.second) do puts "I fire every second, starting in 5 minutes" end
When used inside a WildFly cluster, jobs are highly-available singletons by default. This means that if a job with the same id is scheduled on multiple nodes, it will only initially run on the first node where it fires. If that node goes down, one of the other nodes where that same job is scheduled will automatically take over.
If you want to schedule a job with the same id on all nodes that does actually need to run on every node, you can disable the singleton behavior as part of the schedule for the job:
TorqueBox::Scheduling::Scheduler.schedule("every-node", every: 1000, singleton: false) do puts "I fire on every node where I'm scheduled" end
To unschedule a previously scheduled job, you have two options:
call Scheduler.unschedule with the same
idyou used when scheduling the job:
call Job#unschedule on the job object returned by
job = TorqueBox::Scheduling::Scheduler.schedule(:job1, every: 1000) do puts "I fire every second" end
Unscheduling a job will only unschedule it from the current node in a cluster.