Package bifrost contains functionality to create an in-process job queue with a configurable number of goroutine via workers. It also includes the ability to query scheduled jobs for status (completed jobs are purged at a configurable interval)
package main
import (
"encoding/json"
"fmt"
"github.com/serdmanczyk/bifrost"
"os"
"time"
)
func main() {
stdoutWriter := json.NewEncoder(os.Stdout)
dispatcher := bifrost.NewWorkerDispatcher(
bifrost.Workers(4),
bifrost.JobExpiry(time.Millisecond),
)
// Queue a job func
tracker := dispatcher.QueueFunc(func() error {
time.Sleep(time.Microsecond)
return nil
})
// Queue a 'JobRunner'
dispatcher.Queue(bifrost.JobRunnerFunc(func() error {
time.Sleep(time.Microsecond)
return nil
}))
// Print out incomplete status
status := tracker.Status()
stdoutWriter.Encode(&status)
// {"ID":0,"Complete":false,"Start":"2017-03-23T21:51:27.140681968-07:00"}
// wait on completion
<-tracker.Done()
// Status is now complete
status = tracker.Status()
stdoutWriter.Encode(&status)
// {"ID":0,"Complete":true,"Success":true,"Start":"2017-03-23T21:51:27.140681968-07:00","Finish":"2017-03-23T21:51:27.140830827-07:00"}
// Queue a job that will 'fail'
tracker = dispatcher.QueueFunc(func() error {
time.Sleep(time.Microsecond)
return fmt.Errorf("Failed")
})
// Show failure status
<-tracker.Done()
status = tracker.Status()
stdoutWriter.Encode(&status)
// {"ID":2,"Complete":true,"Success":false,"Error":"Failed","Start":"2017-03-23T21:51:27.141026625-07:00","Finish":"2017-03-23T21:51:27.141079871-07:00"}
// Query for a job's status.
tracker, _ = dispatcher.JobStatus(tracker.ID())
status = tracker.Status()
stdoutWriter.Encode(&status)
// {"ID":2,"Complete":true,"Success":false,"Error":"Failed","Start":"2017-03-23T21:51:27.141026625-07:00","Finish":"2017-03-23T21:51:27.141079871-07:00"}
// Show all jobs
jobs := dispatcher.Jobs()
stdoutWriter.Encode(jobs)
// [{"ID":2,"Complete":true,"Success":false,"Error":"Failed","Start":"2017-03-23T21:51:27.141026625-07:00","Finish":"2017-03-23T21:51:27.141079871-07:00"},{"ID":0,"Complete":true,"Success":true,"Start":"2017-03-23T21:51:27.140681968-07:00","Finish":"2017-03-23T21:51:27.140830827-07:00"},{"ID":1,"Complete":true,"Success":true,"Start":"2017-03-23T21:51:27.140684331-07:00","Finish":"2017-03-23T21:51:27.140873087-07:00"}]
// wait for jobs to be purged
time.Sleep(time.Millisecond * 5)
// should now be empty
jobs = dispatcher.Jobs()
stdoutWriter.Encode(jobs)
// []
dispatcher.Stop()
}If you've read the blog posts Handling 1 Million Requests per Minute with Go or Writing worker queues, in Go this will look very familiar. The main machinery in Bifrost is basically identical to the functionality described in those blog posts, but with a couple added features I wanted for my project.
Added Features:
- Generic jobs: any
func() erroror type that implementsfunc Run() errorcan be queued as a job. - Graceful shutdown: when dispatcher is stopped, waits for running jobs to complete.
- Tracking: queued jobs are given an ID that can be used to query for status later.
- Cleanup: completed jobs are cleaned up after a configurable amount of time.
Lacks (might add these later):
- Lost jobs: if the dispatcher is stopped before all jobs are sent to a worker, unsent jobs may be ignored.
- Errant jobs: jobs taking longer than expected cannot be cancelled.
- Single process: this package does not include functionality to schedule jobs across multiple processes via AMQP, gRPC, or otherwise.
For an example, see the test or example command line app.
Obligatory "not for use in production" but I do welcome feedback.
Bifröst (pronounce B-eye-frost popularly, or traditionally more like Beefroast) is the bridge between the realms of Earth and Asgard (the heavens) in norse mythology.
The Futhark ᛉ Elhaz/Algiz is seen as the symbol for Bifröst, or at least according to this thing I Googled.
The symbology intended is that dispatcher is a 'bridge' between the scheduling goroutine and the worker goroutine.
Honestly I just needed a cool Norse thing to name this, I was reaching. Not to be taken too seriously.