----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/13436/ -----------------------------------------------------------
(Updated Aug. 9, 2013, 5:15 a.m.) Review request for Flume. Changes ------- Added the requested changes. The race condition you mentioned shouldn't be an issue. Even as is the same situation would result in constantly queued resets, so the behavior now is probably preferable. Added some comments. Repository: flume-git Description ------- I removed the lock and set a flag in scheduled runnable. This flag is checked in process and a reconnect is performed if set. Diffs (updated) ----- flume-ng-core/src/main/java/org/apache/flume/sink/AbstractRpcSink.java b3208fc flume-ng-core/src/test/java/org/apache/flume/sink/TestAvroSink.java 8760c25 Diff: https://reviews.apache.org/r/13436/diff/ Testing ------- Unit tests were adjusted(since a reconnect requires process to actually happen) and pass. We tested with and without the patch on our servers and everything looks good with log duplication at the same level as having interval = 0 Thanks, Juhani Connolly
