Class: Karafka::Pro::Processing::Schedulers::Base

Inherits:
Object
  • Object
show all
Defined in:
lib/karafka/pro/processing/schedulers/base.rb

Overview

Note:

All the on_ methods can be redefined with a non-thread-safe versions without locks if needed, however when doing so, ensure that your scheduler is stateless.

Base for all the Pro custom schedulers

It wraps the Scheduler API with mutex to ensure, that during scheduling we do not start scheduling other work that could impact the decision making in between multiple subscription groups running in separate threads.

Direct Known Subclasses

Default

Instance Method Summary collapse

Constructor Details

#initialize(queue) ⇒ Base

Returns a new instance of Base.

Parameters:



29
30
31
32
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 29

def initialize(queue)
  @queue = queue
  @mutex = Mutex.new
end

Instance Method Details

#clear(_group_id) ⇒ Object

By default schedulers are stateless, so nothing to clear.

Parameters:

  • _group_id (String) —

    Subscription group id



130
131
132
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 130

def clear(_group_id)
  nil
end

#manage ⇒ Object

Should manage scheduling on jobs state changes

By default does nothing as default schedulers are stateless



116
117
118
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 116

def manage
  nil
end

#on_clear(group_id) ⇒ Object

Runs clearing under mutex

Parameters:

  • group_id (String) —

    Subscription group id



123
124
125
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 123

def on_clear(group_id)
  @mutex.synchronize { clear(group_id) }
end

#on_manage ⇒ Object

Runs the manage tick under mutex



109
110
111
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 109

def on_manage
  @mutex.synchronize { manage }
end

#on_schedule_consumption(jobs_array) ⇒ Object

Runs the consumption jobs scheduling flow under a mutex

Parameters:

  • jobs_array —

    [Array<Karafka::Processing::Jobs::Consume, Processing::Jobs::ConsumeNonBlocking>] jobs for scheduling



47
48
49
50
51
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 47

def on_schedule_consumption(jobs_array)
  @mutex.synchronize do
    schedule_consumption(jobs_array)
  end
end

#on_schedule_idle(jobs_array) ⇒ Object

Runs the idle jobs scheduling flow under a mutex

Parameters:



85
86
87
88
89
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 85

def on_schedule_idle(jobs_array)
  @mutex.synchronize do
    schedule_idle(jobs_array)
  end
end

#on_schedule_periodic(jobs_array) ⇒ Object

Runs the periodic jobs scheduling flow under a mutex

Parameters:

  • jobs_array —

    [Array<Processing::Jobs::Periodic, Processing::Jobs::PeriodicNonBlocking>] jobs for scheduling



96
97
98
99
100
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 96

def on_schedule_periodic(jobs_array)
  @mutex.synchronize do
    schedule_periodic(jobs_array)
  end
end

#on_schedule_revocation(jobs_array) ⇒ Object

Runs the revocation jobs scheduling flow under a mutex

Parameters:

  • jobs_array —

    [Array<Karafka::Processing::Jobs::Revoked, Processing::Jobs::RevokedNonBlocking>] jobs for scheduling



67
68
69
70
71
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 67

def on_schedule_revocation(jobs_array)
  @mutex.synchronize do
    schedule_revocation(jobs_array)
  end
end

#on_schedule_shutdown(jobs_array) ⇒ Object

Runs the shutdown jobs scheduling flow under a mutex

Parameters:



76
77
78
79
80
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 76

def on_schedule_shutdown(jobs_array)
  @mutex.synchronize do
    schedule_shutdown(jobs_array)
  end
end

#schedule_consumption(_jobs_array) ⇒ Object

Should schedule the consumption jobs

Parameters:

  • _jobs_array —

    [Array<Karafka::Processing::Jobs::Consume, Processing::Jobs::ConsumeNonBlocking>] jobs for scheduling

Raises:

  • (NotImplementedError) 


58
59
60
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 58

def schedule_consumption(_jobs_array)
  raise NotImplementedError, 'Implement in a subclass'
end

#schedule_fifo(jobs_array) ⇒ Object Also known as: schedule_revocation, schedule_shutdown, schedule_idle, schedule_periodic

Schedules any jobs provided in a fifo order

Parameters:



36
37
38
39
40
 
# File 'lib/karafka/pro/processing/schedulers/base.rb', line 36

def schedule_fifo(jobs_array)
  jobs_array.each do |job|
    @queue << job
  end
end