I just thought may be we could put a warning whenever that error comes user
can tune either memoryFraction or executor memory options. And this warning
get's displayed when TaskSetManager receives task failures due to OOM.
Prashant Sharma
On Mon, May 5, 2014 at 2:10 PM, Ajay Nair prodig...@gmail.com wrote:
Hi,
I am using 1 master and 3 slave workers for processing 27gb of Wikipedia
data that is tab separated and every line contains wikipedia page
information. The tab separated data has title of the page and the page
contents. I am using the regular expression to extract links as mentioned
in
the site below:
http://ampcamp.berkeley.edu/big-data-mini-course/graph-analytics-with-graphx.html#running-pagerank-on-wikipedia
Although it runs fne for around 300Mb data set, it runs in to issues when I
try to execute the same code using the 27gb data from hdfs.
The error thrown is given below:
14/05/05 07:15:22 WARN scheduler.TaskSetManager: Loss was due to
java.lang.OutOfMemoryError
java.lang.OutOfMemoryError: GC overhead limit exceeded
at java.util.regex.Matcher.init(Matcher.java:224)
Is there any way to over come this issue?
My cluster is a ec2 m3.large machine.
Thanks
Ajay
--
View this message in context:
http://apache-spark-developers-list.1001551.n3.nabble.com/Apache-spark-on-27gb-wikipedia-data-tp6487.html
Sent from the Apache Spark Developers List mailing list archive at
Nabble.com.