We should not do a checkpoint if we do't have sufficient free space. The same approach as FLUME-1609 can be taken.
FileChannel check for full disks in the background
FileChannel detecting when the underlying file systems are full could provide cleaner error recovery
Integrated in flume-trunk #339 (See https://builds.apache.org/job/flume-trunk/339/)
FLUME-1763. FileChannel checkpoints should not be done without free space (Revision 806a7b6ace9398de2645f073acc72de19913bfab)
Result = SUCCESS
hshreedharan : http://git-wip-us.apache.org/repos/asf/flume/repo?p=flume.git&a=commit&h=806a7b6ace9398de2645f073acc72de19913bfab
Patch committed, rev: 806a7b6ace9398de2645f073acc72de19913bfab. Thanks Brock!
Latest patch attached.
Note that although the checkpoint file itself is pre-allocated, the meta data file is not.
We should also:
1) Make that number configurable.
2) Increase the default to 100MB