If not set explicitly with spark.default.parallelism, it will default to the number of cores currently available (minimum 2). At the very start, some executors haven't completed registering, which I think explains why it goes up after a short time. (In the case of dynamic allocation it will change over time.) You can set it explicitly to match what you set the executor count and cores to.
On Mon, Jul 6, 2020 at 10:35 PM Stephen Coy <s...@infomedia.com.au.invalid> wrote: > > Hi there, > > I have found that if I invoke > > sparkContext.defaultParallelism() > > too early it will not return the correct value; > > For example, if I write this: > > final JavaSparkContext sparkContext = new > JavaSparkContext(sparkSession.sparkContext()); > final int workerCount = sparkContext.defaultParallelism(); > > I will get some small number (which I can’t recall right now). > > However, if I insert: > > sparkContext.parallelize(List.of(1, 2, 3, 4)).collect() > > between these two lines I get the expected value being something like > node_count * node_core_count; > > This seems like a hacky work around solution to me. Is there a better way to > get this value initialised properly? > > FWIW, I need this value to size a connection pool (fs.s3a.connection.maximum) > correctly in a cluster independent way. > > Thanks, > > Steve C > > > [http://downloads.ifmsystems.com/data/marketing/images/signatures/driving-force-newsletter.jpg]<https://www.infomedia.com.au/driving-force/?utm_campaign=200630%20Email%20Signature&utm_source=Internal&utm_medium=Email&utm_content=Driving%20Force> > This email contains confidential information of and is the copyright of > Infomedia. It must not be forwarded, amended or disclosed without consent of > the sender. If you received this message by mistake, please advise the sender > and delete all copies. Security of transmission on the internet cannot be > guaranteed, could be infected, intercepted, or corrupted and you should > ensure you have suitable antivirus protection in place. By sending us your or > any third party personal details, you consent to (or confirm you have > obtained consent from such third parties) to Infomedia’s privacy policy. > http://www.infomedia.com.au/privacy-policy/ --------------------------------------------------------------------- To unsubscribe e-mail: user-unsubscr...@spark.apache.org
