Status

Current state: Under DiscussionAccepted

Discussion thread: here

JIRA: here [Change the link from KAFKA-1 to your own ticket]

Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).

...

If the automatic partitioning is needed (i.e. no record partition or key is specified; or partitioner.ignore.keys is set to true):

1. If the client rack is specified and there are partition leaders in the same rack, select the next partition from those “nearby” partitions. The current algorithm is followed in the other aspects, including load balancing.
2. Select the next partition from all partitions following the current algorithm in the following cases:
   1. if the client rack is not specified;
   2. if there are no partitions with leaders in the same rack as the client;
   3. if all the “nearby” partitions are unavailable.
   4. if partitioner.rack.aware is false.

...