v0.4.0
This release integrates an exponential backoff procedure into all of the consumer's internal processing loops so that they will not attempt to poll Redis in tight loops if they encounter network failures. It also exposes the ability to automatically trigger a critical error and abort after a maximum number of such failures. The default behavior is still for the processing loops to attempt to reconnect indefinitely.
Other changes:
- The maximum backoff between job retries is now configurable. The default parameters for retries have been modified so that jobs will now be retried for roughly 2 weeks before being killed.
DefaultLogger
is now less verbose by default. ALoudLogger
type has been added for easy debugging.- A race condition was fixed where consumers might fail to register themselves before retrieving jobs in cases of intermittent network failure.
- Various performance enhancements and increased test coverage.
This is expected to be the final feature release before v1. Future pre-releases will focus on bug fixes, usability enhancements, and documentation improvements.