Hi,
Thanks for the prompt reply.
I checked the code. The main issue is the large number of mappers. If the
number of mappers is set to some number around 1000, there will be no
problem. I hope the bug gets fixed in the next releases.
On Mon, Jan 5, 2015 at 1:26 AM, Josh Rosen
Ah, so I guess this *is* still an issue since we needed to use a bitmap for
tracking zero-sized blocks (see
https://issues.apache.org/jira/browse/SPARK-3740; this isn't just a
performance issue; it's necessary for correctness). This will require a
bit more effort to fix, since we'll either have
Which version of Spark are you using? It seems like the issue here is that
the map output statuses are too large to fit in the Akka frame size. This
issue has been fixed in Spark 1.2 by using a different encoding for map
outputs for jobs with many reducers (
I use the 1.2 version.
On Sun, Jan 4, 2015 at 3:01 AM, Josh Rosen rosenvi...@gmail.com wrote:
Which version of Spark are you using? It seems like the issue here is
that the map output statuses are too large to fit in the Akka frame size.
This issue has been fixed in Spark 1.2 by using a