mirror of https://github.com/apache/druid.git
cb03efeb14
* Kafka Index Task that supports Incremental handoffs - Incrementally handoff segments when they hit maxRowsPerSegment limit - Decouple segment partitioning from Kafka partitioning, all records from consumed partitions go to a single druid segment - Support for restoring task on middle manager restarts by check pointing end offsets for segments * take care of review comments * make getCurrentOffsets call async, keep track of publishing sequence, review comments * fix setEndoffset duplicate request handling, formatting * fix unit test * backward compatibility * make AppenderatorDriverMetadata backwards compatible * add unit test * fix deadlock between persist and push executors in AppenderatorImpl * fix formatting * use persist dir instead of work dir * review comments * fix deadlock * actually fix deadlock |
||
|---|---|---|
| .. | ||
| src | ||
| pom.xml | ||