Class: Sidekiq::Fetcher

Inherits:
Object
  • Object
show all
Includes:
Celluloid, Util
Defined in:
lib/sidekiq/fetch.rb

Overview

The Fetcher blocks on Redis, waiting for a message to process from the queues. It gets the message and hands it to the Manager to assign to a ready Processor.

Constant Summary collapse

TIMEOUT = 
1

Constants included from Util

Util::EXPIRY

Class Method Summary collapse

Instance Method Summary collapse

Methods included from Util

#hostname, #logger, #process_id, #redis, #watchdog

Methods included from ExceptionHandler

#handle_exception

Constructor Details

#initialize(mgr, options) ⇒ Fetcher

Returns a new instance of Fetcher.



15
16
17
18
 
# File 'lib/sidekiq/fetch.rb', line 15

def initialize(mgr, options)
  @mgr = mgr
  @strategy = Fetcher.strategy.new(options)
end

Class Method Details

.done! ⇒ Object

Ugh. Say hello to a bloody hack. Can’t find a clean way to get the fetcher to just stop processing its mailbox when shutdown starts.



52
53
54
 
# File 'lib/sidekiq/fetch.rb', line 52

def self.done!
  @done = true
end

.done? ⇒ Boolean

Returns:

  • (Boolean) 


56
57
58
 
# File 'lib/sidekiq/fetch.rb', line 56

def self.done?
  @done
end

.strategy ⇒ Object 



60
61
62
 
# File 'lib/sidekiq/fetch.rb', line 60

def self.strategy
  Sidekiq.options[:fetch] || BasicFetch
end

Instance Method Details

#fetch ⇒ Object

Fetching is straightforward: the Manager makes a fetch request for each idle processor when Sidekiq starts and then issues a new fetch request every time a Processor finishes a message.

Because we have to shut down cleanly, we can’t block forever and we can’t loop forever. Instead we reschedule a new fetch if the current fetch turned up nothing.



28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
 
# File 'lib/sidekiq/fetch.rb', line 28

def fetch
  watchdog('Fetcher#fetch died') do
    return if Sidekiq::Fetcher.done?

    begin
      work = @strategy.retrieve_work

      if work
        @mgr.async.assign(work)
      else
        after(0) { fetch }
      end
    rescue => ex
      logger.error("Error fetching message: #{ex}")
      logger.error(ex.backtrace.first)
      sleep(TIMEOUT)
      after(0) { fetch }
    end
  end
end