----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/24992/#review51327 -----------------------------------------------------------
clients/src/main/java/org/apache/kafka/common/requests/MetadataResponse.java <https://reviews.apache.org/r/24992/#comment89517> Would it be simpler to just don't do the test on partError? The rest of the code can deal with a null leader already. It's probably a bit weird to ignore a partition whose leader is not available. For example, if a message has to go to a partition (e.g., keyed message), we want to know that partition whether the leader is available or not. - Jun Rao On Aug. 22, 2014, 8:04 p.m., Dong Lin wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/24992/ > ----------------------------------------------------------- > > (Updated Aug. 22, 2014, 8:04 p.m.) > > > Review request for kafka. > > > Bugs: KAFKA-1609 > https://issues.apache.org/jira/browse/KAFKA-1609 > > > Repository: kafka > > > Description > ------- > > KAFKA-1609; New producer metadata response handling should only exclude a > PartitionInfo when its error is LEADER_NOT_AVAILABLE > > > Diffs > ----- > > > clients/src/main/java/org/apache/kafka/common/requests/MetadataResponse.java > 7d90fceffd4a92460c616f8bacd78057a38a3c16 > > Diff: https://reviews.apache.org/r/24992/diff/ > > > Testing > ------- > > > Thanks, > > Dong Lin > >
