Module: Karafka::Pro::Processing::Strategies::Dlq::Mom
- Includes:
- Default
- Included in:
- Aj::DlqMom, MomVp
- Defined in:
- lib/karafka/pro/processing/strategies/dlq/mom.rb
Overview
Strategy for supporting DLQ with Mom enabled
Constant Summary collapse
- FEATURES =
Features for this strategy
%i[ dead_letter_queue manual_offset_management ].freeze
Instance Method Summary collapse
-
#handle_after_consume ⇒ Object
When manual offset management is on, we do not mark anything as consumed automatically and we rely on the user to figure things out.
Methods included from Default
#build_dlq_message, #dispatch_to_dlq, #dispatch_to_dlq?, #find_skippable_message
Methods included from Karafka::Pro::Processing::Strategies::Default
#handle_before_consume, #handle_before_enqueue, #handle_consume, #handle_revoked
Methods included from Karafka::Processing::Strategies::Default
#commit_offsets, #commit_offsets!, #handle_before_consume, #handle_before_enqueue, #handle_consume, #handle_idle, #handle_revoked, #handle_shutdown, #mark_as_consumed, #mark_as_consumed!
Methods included from Karafka::Processing::Strategies::Base
#handle_before_consume, #handle_before_enqueue, #handle_consume, #handle_idle, #handle_revoked, #handle_shutdown
Instance Method Details
#handle_after_consume ⇒ Object
When manual offset management is on, we do not mark anything as consumed automatically and we rely on the user to figure things out
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 |
# File 'lib/karafka/pro/processing/strategies/dlq/mom.rb', line 32 def handle_after_consume coordinator.on_finished do return if revoked? if coordinator.success? coordinator.pause_tracker.reset elsif coordinator.pause_tracker.attempt <= topic.dead_letter_queue.max_retries retry_after_pause # If we've reached number of retries that we could, we need to skip the first # message that was not marked as consumed, pause and continue, while also moving # this message to the dead topic. # # For a Mom setup, this means, that user has to manage the checkpointing by # himself. If no checkpointing is ever done, we end up with an endless loop. else # We reset the pause to indicate we will now consider it as "ok". coordinator.pause_tracker.reset , = dispatch_to_dlq() if dispatch_to_dlq? # Save the next offset we want to go with after moving given message to DLQ # Without this, we would not be able to move forward and we would end up # in an infinite loop trying to un-pause from the message we've already processed # Of course, since it's a MoM a rebalance or kill, will move it back as no # offsets are being committed coordinator.seek_offset = .offset + 1 pause(coordinator.seek_offset, nil, false) end end end |