A Sidekiq-compatible background worker system, built on Keenser and using Redis for job persistence

Usage

See the demo for an example setup

Define Your Jobs

In e.g. Jobs.hs

import Yesod.Worker

blah :: Worker Handler (Text, Int)
-- Worker name (should be unique), queue name, then arg list
blah = Worker "blah" "default" $ \(word, n) ->
  -- Your long-running work goes here
  -- N.B. this runs in the Yesod Handler monad, but doesn't have access to
  --   anything request- or response- related. Under the hood, it's using
  --   `unsafeHandler`, and may error if you try to e.g. redirect. A future
  --   version will include a `Handler`-like monad stack which should be
  --   safe from these sorts of errors.
  replicateM_ n $ do
    $(logWarn) word
    liftIO . threadDelay $ 1000000

setup = do
  register blah
  -- register any other jobs you like
  -- they need not have the same argument type

Include the Subsite

In Foundation.hs

import Yesod.Worker

data App = App
  { ...
  , getWorkers :: Workers
  }

instance YesodWorker App where
  workers = getWorkers

In Application.hs

import Yesod.Worker
import Jobs

makeFoundation = do
  ...
  getWorkers <- newWorkers
  ...

makeApplication foundation = do
  unsafeHandler foundation $ do
    bootWorkers setup
    -- optionally, queue jobs on app boot:
    enqueue blah ("queued at start", 2)

Queue Up a Job

In e.g. Handlers/Whatever.hs

import Yesod.Worker
import Jobs

postHandlerR = do 
  enqueue blah ("na", 16)
  ...

Watch the Stats

Add a route

/workers WorkerR Workers getWorkers

and visit /workers

Over time, this page should expand to a Sidekiq-style administration panel, allowing for closer monitoring and options for retrying from the web UI.

Note that the Redis interface is already Sidekiq compatable, so you can literally use sidekiq/web for monitoring as well.