> On June 25, 2012, 7:47 p.m., Brock Noland wrote: > > /trunk/flume-ng-channels/flume-file-channel/src/main/java/org/apache/flume/channel/file/FlumeEventQueue.java, > > line 97 > > <https://reviews.apache.org/r/5556/diff/1/?file=116345#file116345line97> > > > > If someone kept the default heap as is (20MB) but raise capacity to 2 > > million (2M * 8 bytes = 15MB) I think this would fail. The capacity could > > be 2M with a heap of 20MB since the queue itself is kept off heap.
Thanks for the feedback Brock, this makes sense. I will break it up into chunks of 2MBs and test. - Arvind ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/5556/#review8563 ----------------------------------------------------------- On June 25, 2012, 7:05 p.m., Arvind Prabhakar wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/5556/ > ----------------------------------------------------------- > > (Updated June 25, 2012, 7:05 p.m.) > > > Review request for Flume and Brock Noland. > > > Description > ------- > > Modified the checkpoint allocation to write a byte array in order to allocate > the capacity. > > > This addresses bug FLUME-1304. > https://issues.apache.org/jira/browse/FLUME-1304 > > > Diffs > ----- > > > /trunk/flume-ng-channels/flume-file-channel/src/main/java/org/apache/flume/channel/file/FlumeEventQueue.java > 1353634 > > Diff: https://reviews.apache.org/r/5556/diff/ > > > Testing > ------- > > Ran all tests. Manual verification of file allocation using byte array > indicates much better performance than the previous implementation. > > > Thanks, > > Arvind Prabhakar > >
