My settings are: Running Spark 2.1 on 3 node YARN cluster with 160 GB.
Dynamic allocation turned on. spark.executor.memory=6G,
spark.executor.cores=6
First, I am reading hive tables: orders(329MB) and lineitems(1.43GB) and
doing left outer join.
Next, I apply 7 different filter conditions based on joined
dataset(something line var line1=joinedDf.filter("linenumber=1"),var
line2=joinedDf.filter("l_linenumber=2, etc). Because I'm doing filter on
joned dataset multiple times, I thought doing a persist(MEMORY_ONLY)
should
help here as the joined dataset will fit fully in memory.
1. I noticed that with persist, spark job takes longer time to run than
non-persist(3.5 mins vs 3.3 mins). With persist, the DAG shows that a
single
stage got created for persist and other downstream jobs are waiting for
the
persist to complete. Does that mean persist is a blocking call? Or do
stages
in other jobs start processing as and when persisted blocks become
available?
2. In non-persist case, different jobs are creating different stages to
read
the same data. Data is read multiple times in different stages, but this
is
still is turning out to be faster than the persist case.
3. With larger data sets, persist actually causes executors to run out
of
memory: Java heap space. Without persist, the spark jobs complete just
fine.
I looked at some other suggestions here: Spark
java.lang.OutOfMemoryError:
Java heap space I tried increasing/decreasing executor cores,
persisting
with disk only, increasing partitions, modifying storage ratio, but
nothing
seems to help with executor memory issues.
4. Also, I posted this on stack overflow and tried those suggestions of
persisting joined dataset and doing a count before apply filter conditions,
but that did not improve persist performance:
https://stackoverflow.com/questions/46101585/persist-slower-than-non-persist-calls
Would appreciate if someone could mention how persist works, in what
cases
it is faster than not-persisting and more importantly, how to go about
troubleshooting out of memory issues.
--
Sent from: http://apache-spark-user-list.1001560.n3.nabble.com/
---------------------------------------------------------------------
To unsubscribe e-mail: [email protected]
