Skip to content

[Event Hubs] GeoDR Failover Error Handling #47303

@jsquire

Description

@jsquire

Problem Statement

When a GeoDR failover happens, it is possible that a legacy checkpoint exists in a format that is not GeoDR-aware. In this case, when the processor attempts to recover from the failover, the Event Hubs service will reject the reader with a unique error that indicates the checkpoint is invalid and cannot be used.

Scope of Work

  • Follow-up with the Event Hubs service team for error details.

  • Update EventProcessor<T> such that if a reader fails when opening a link with the GeoDR error condition, it recognizes the scenario and can take steps to recover.

Success Criteria

  • When the service returns a GeoDR error condition when the processor is creating a partition reader, that error is identified and has special handling.
  • The processor logic has been updated to ignore/remove the invalid checkpoint when the GeoDR error scenario takes place.
  • The processor is resilient and can resume operations after a failover, whether checkpoints are valid or invalid.
  • The tests needed to validate functionality have been added and pass reliably.
  • The existing tests continue to pass reliably.

Metadata

Metadata

Assignees

Labels

ClientThis issue is related to a non-management packageEvent Hubs

Type

No type

Projects

Status

Done

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions