Class: Karafka::Pro::RecurringTasks::Consumer

Inherits:
BaseConsumer
  • Object
show all
Defined in:
lib/karafka/pro/recurring_tasks/consumer.rb

Overview

Consumer responsible for management of the recurring tasks and their execution There are some assumptions made here that always need to be satisfied: - we only run schedules that are of same or newer version - we always mark as consumed in such a way, that the first message received after assignment (if any) is a state

Instance Attribute Summary

Attributes inherited from BaseConsumer

#client, #coordinator, #id, #messages, #producer

Instance Method Summary collapse

Methods inherited from BaseConsumer

#on_after_consume, #on_before_consume, #on_before_schedule_consume, #on_before_schedule_eofed, #on_before_schedule_idle, #on_before_schedule_revoked, #on_before_schedule_shutdown, #on_consume, #on_eofed, #on_idle, #on_initialized, #on_revoked, #on_shutdown

Constructor Details

#initialize(*args) ⇒ Consumer

Returns a new instance of Consumer.

Parameters:

  • args (Array) —

    all arguments accepted by the consumer



24
25
26
27
 
# File 'lib/karafka/pro/recurring_tasks/consumer.rb', line 24

def initialize(*args)
  super
  @executor = Executor.new
end

Instance Method Details

#consume ⇒ Object

Raises:



29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
 
# File 'lib/karafka/pro/recurring_tasks/consumer.rb', line 29

def consume
  # There is nothing we can do if we operate on a newer schedule. In such cases we should
  # just wait and re-raise error hoping someone will notice or that this will be
  # reassigned to a process with newer schedule
  raise(Errors::IncompatibleScheduleError) if @executor.incompatible?

  messages.each do |message|
    payload = message.payload
    type = payload[:type]

    case type
    when 'schedule'
      # If we're replaying data, we need to record the most recent stored state, so we
      # can use this data to fully initialize the scheduler
      @executor.update_state(payload) if @executor.replaying?

      # If this is first message we cannot mark it on the previous offset
      next if message.offset.zero?

      # We always mark as consumed in such a way, that when replaying, we start from a
      # schedule state message. This makes it easier to recover.
      mark_as_consumed Karafka::Messages::Seek.new(
        topic.name,
        partition,
        message.offset - 1
      )
    when 'command'
      @executor.apply_command(payload)

      next if @executor.replaying?

      # Execute on each incoming command to have nice latency but only after replaying
      # During replaying we should not execute because there may be more state changes
      # that collectively have a different outcome
      @executor.call
    else
      raise ::Karafka::Errors::UnsupportedCaseError, type
    end
  end

  eofed if eofed?
end

#eofed ⇒ Object

Starts the final replay process if we reached eof during replaying



73
74
75
76
77
78
79
 
# File 'lib/karafka/pro/recurring_tasks/consumer.rb', line 73

def eofed
  # We only mark as replayed if we were replaying in the first place
  # If already replayed, nothing to do
  return unless @executor.replaying?

  @executor.replay
end

#tick ⇒ Object

Runs the cron execution if all good.



82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
 
# File 'lib/karafka/pro/recurring_tasks/consumer.rb', line 82

def tick
  # Do nothing until we fully recover the correct state
  return if @executor.replaying?

  # If the state is incompatible, we can only raise an error.
  # We do it here and in the `#consume` so the pause-retry kicks in basically reporting
  # on this issue once every minute. That way user should not miss this.
  # We use seek to move so we can achieve a pause of 60 seconds in between consecutive
  # errors instead of on each tick because it is much more frequent.
  if @executor.incompatible?
    if messages.empty?
      raise Errors::IncompatibleScheduleError
    else
      return seek(messages.last.offset - 1)
    end
  end

  # If all good and compatible we can execute the recurring tasks
  @executor.call
end