Thanks Eric!
Sent from my iPhone
On Feb 23, 2010, at 8:02 PM, Eric Yang <ey...@yahoo-inc.com> wrote:
This is because Chukwa doesn't use HDFS append feature. Any data that
arrived late, they will be labeled as .#.evt. In this case, it
looks like
there were two spills of data after daily rolling is done. It would
be nice
to have a scheduler to repack the data again after spill, but this
feature
has not been implemented.
Regards,
Eric
On 2/23/10 2:20 PM, "Corbin Hoenes" <cor...@tynt.com> wrote:
We are seeing files like this after the rotation jobs have run:
MyFile_20100222.1.evt (150 MB)
MyFile_20100222.2.evt (5 MB)
MyFile_DailyDone_20100222.1.evt (64 GB)
What are the smaller files for?
