Skip to content

logpuller: simplify region failure handling pipeline #5612

Description

@lidezhu

The current region failure path has an extra dispatch stage between buffering failed regions and handling them.

Today the flow is:

  • append failed regions into errCache
  • dispatch cached items into errCh
  • read errCh and run failure handling / retry logic

This extra stage adds one more channel hop and one more goroutine to the recovery path, even though the final retry and reschedule logic stays the same.

We can simplify this by letting the failure handler consume batches directly from errCache, instead of forwarding them through errCh first.

Metadata

Metadata

Assignees

Labels

affects-8.5This bug affects the 8.5.x(LTS) versions.type/enhancementThe issue or PR belongs to an enhancement.

Type

No type

Fields

No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions