spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Gabor Feher (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (SPARK-15796) Reduce spark.memory.fraction default to avoid overrunning old gen in JVM default config
Date Sun, 12 Jun 2016 18:27:21 GMT

     [ https://issues.apache.org/jira/browse/SPARK-15796?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Gabor Feher updated SPARK-15796:
--------------------------------
    Attachment: baseline.txt
                memfrac06.txt
                memfrac063.txt
                memfrac066.txt

I've run the above code in four different setup:

memfrac06.txt; spark.memory.fraction=0.6
total time= 45s, GC time= 2s
memfrac063.txt; spark.memory.fraction=0.63
total time= 56s, GC time= 11s
memfrac066.txt; spark.memory.fraction=0.66
total time= 270s, GC time= 225s
baseline.txt; spark.memory.fraction=0.75 (default)
total time= 270s, GC time= 233s

(Note that numbers are from single runs, so there is some noise.)
Also see that, spark.memory.fraction=0.66 is not low enough to fix the problem.

Here is an example line from the output of the baseline case (spark.memory.fraction=0.75):
[Full GC [PSYoungGen: 350208K->0K(699392K)] [ParOldGen: 2028768K->2028681K(2097152K)]
2378976K->2028681K(2796544K) [PSPermGen: 44023K->44023K(91136K)], 1.9907560 secs] [Times:
user=6.62 sys=0.02, real=1.99 secs] 

See that the old generation is nearly full. I think if it's over a threshold close to full,
then full GC gets triggered and that causes the slowdown. This would explain why using exactly
0.66 as a threshold is not fixing the issue, but lower threshold do.

p.s.: I've seen [~andrewor14]'s presentation at the Spark Summit:
https://spark-summit.org/2016/speakers/andrew-or/
I think he has mentioned that one difference between storage and execution memory is that
execution is short-lived while storage is long-lived. So one idea can be to introduce a new
config value, e.g. spark.memory.cacheUpperLimit which is always lower than the old generation
ratio. But at the same time, keep spark.memory.fraction at a higher level to allow for more
execution memory. I am not sure if this would work well, but it might be one thing to try
and measure its performance.

> Reduce spark.memory.fraction default to avoid overrunning old gen in JVM default config
> ---------------------------------------------------------------------------------------
>
>                 Key: SPARK-15796
>                 URL: https://issues.apache.org/jira/browse/SPARK-15796
>             Project: Spark
>          Issue Type: Improvement
>    Affects Versions: 1.6.0, 1.6.1
>            Reporter: Gabor Feher
>            Priority: Minor
>         Attachments: baseline.txt, memfrac06.txt, memfrac063.txt, memfrac066.txt
>
>
> While debugging performance issues in a Spark program, I've found a simple way to slow
down Spark 1.6 significantly by filling the RDD memory cache. This seems to be a regression,
because setting "spark.memory.useLegacyMode=true" fixes the problem. Here is a repro that
is just a simple program that fills the memory cache of Spark using a MEMORY_ONLY cached RDD
(but of course this comes up in more complex situations, too):
> {code}
> import org.apache.spark.SparkContext
> import org.apache.spark.SparkConf
> import org.apache.spark.storage.StorageLevel
> object CacheDemoApp { 
>   def main(args: Array[String]) {
>     val conf = new SparkConf().setAppName("Cache Demo Application")                 
                     
>     val sc = new SparkContext(conf)
>     val startTime = System.currentTimeMillis()
>                                                                                     
                     
>     val cacheFiller = sc.parallelize(1 to 500000000, 1000)                          
                     
>       .mapPartitionsWithIndex {
>         case (ix, it) =>
>           println(s"CREATE DATA PARTITION ${ix}")                                   
                     
>           val r = new scala.util.Random(ix)
>           it.map(x => (r.nextLong, r.nextLong))
>       }
>     cacheFiller.persist(StorageLevel.MEMORY_ONLY)
>     cacheFiller.foreach(identity)
>     val finishTime = System.currentTimeMillis()
>     val elapsedTime = (finishTime - startTime) / 1000
>     println(s"TIME= $elapsedTime s")
>   }
> }
> {code}
> If I call it the following way, it completes in around 5 minutes on my Laptop, while
often stopping for slow Full GC cycles. I can also see with jvisualvm (Visual GC plugin) that
the old generation of JVM is 96.8% filled.
> {code}
> sbt package
> ~/spark-1.6.0/bin/spark-submit \
>   --class "CacheDemoApp" \
>   --master "local[2]" \
>   --driver-memory 3g \
>   --driver-java-options "-XX:+PrintGCDetails" \
>   target/scala-2.10/simple-project_2.10-1.0.jar
> {code}
> If I add any one of the below flags, then the run-time drops to around 40-50 seconds
and the difference is coming from the drop in GC times:
>   --conf "spark.memory.fraction=0.6"
> OR
>   --conf "spark.memory.useLegacyMode=true"
> OR
>   --driver-java-options "-XX:NewRatio=3"
> All the other cache types except for DISK_ONLY produce similar symptoms. It looks like
that the problem is that the amount of data Spark wants to store long-term ends up being larger
than the old generation size in the JVM and this triggers Full GC repeatedly.
> I did some research:
> * In Spark 1.6, spark.memory.fraction is the upper limit on cache size. It defaults to
0.75.
> * In Spark 1.5, spark.storage.memoryFraction is the upper limit in cache size. It defaults
to 0.6 and...
> * http://spark.apache.org/docs/1.5.2/configuration.html even says that it shouldn't be
bigger than the size of the old generation.
> * On the other hand, OpenJDK's default NewRatio is 2, which means an old generation size
of 66%. Hence the default value in Spark 1.6 contradicts this advice.
> http://spark.apache.org/docs/1.6.1/tuning.html recommends that if the old generation
is running close to full, then setting spark.memory.storageFraction to a lower value should
help. I have tried with spark.memory.storageFraction=0.1, but it still doesn't fix the issue.
This is not a surprise: http://spark.apache.org/docs/1.6.1/configuration.html explains that
storageFraction is not an upper-limit but a lower limit-like thing on the size of Spark's
cache. The real upper limit is spark.memory.fraction.
> To sum up my questions/issues:
> * At least http://spark.apache.org/docs/1.6.1/tuning.html should be fixed. Maybe the
old generation size should also be mentioned in configuration.html near spark.memory.fraction.
> * Is it a goal for Spark to support heavy caching with default parameters and without
GC breakdown? If so, then better default values are needed.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org


Mime
View raw message