ibm bluemix spark submit

Question

I'm new to Bluemix. I have created the Apache Spark Service and I tried to submit a simple hello-world jar through spark submit. (I used this link to fallow: https://console.ng.bluemix.net/docs/services/AnalyticsforApacheSpark/index-gentopic3.html#genTopProcId4

After submitting the jar, the log file says:

Submit job result: { "action" : "CreateSubmissionResponse",
"message" : "Driver successfully submitted as driver-20170602xxxxxxxxxxx", "serverSparkVersion" : "2.0.2",
"submissionId" : "driver-20170602xxxxxxxxxxx", "success" : true }

Submission ID: driver-20170602xxxxxxxxxxx Job submitted : driver-driver-20170602xxxxxxxxxxx Polling job status. Poll #1. Getting status ==== Failed Status output ===================================================== ERROR: Job failed. spark-submit log file: spark-submit_1496403637550663675.log View job's stdout log at stdout_1496403637550663675 View job's stderr log at stderr_1496403637550663675

What may be the problem in this case?

Also I see people talking about notebook and Jupiter for accessing/running spark job. But I don't see the notebook /Jupiter option on my Dashboard.

Thank you for your inputs

The curl on stdout ie,( https://spark.eu-gb.bluemix.net/tenant/data/workdir/driver-20170614074046xxxxxxxxx277e6a/stdout ) is: "no extra configuration defined"

But i found the below error message on stderr:

log4j:ERROR Could not find value for key log4j.appender.FILE log4j:ERROR Could not instantiate appender named "FILE". ERROR deploy.ego.EGOClusterDriverWrapper: Uncaught exception: java.nio.file.NoSuchFileException: /gpfs/fs01/user/sd74-836f4292ca6442xxxxxxxx/data/e717e66fe44f5a1ea7eec81cbd/hellospark_2.11-1.0.jar at sun.nio.fs.UnixException.translateToIOException(UnixException.java:98) at sun.nio.fs.UnixException.rethrowAsIOException(UnixException.java:114) at sun.nio.fs.UnixException.rethrowAsIOException(UnixException.java:119) at sun.nio.fs.UnixCopyFile.copy(UnixCopyFile.java:538) at sun.nio.fs.UnixFileSystemProvider.copy(UnixFileSystemProvider.java:265) at java.nio.file.Files.copy(Files.java:1285) at org.apache.spark.util.Utils$.org$apache$spark$util$Utils$$copyRecursive(Utils.scala:629) at org.apache.spark.util.Utils$.copyFile(Utils.scala:600) at org.apache.spark.util.Utils$.doFetchFile(Utils.scala:685) at org.apache.spark.util.Utils$.fetchFile(Utils.scala:484) at org.apache.spark.deploy.ego.EGOClusterDriverWrapper$$anonfun$startUserClass$2.apply(EGOClusterDriverWrapper.scala:411) at org.apache.sp ark.deploy.ego.EGOClusterDriverWrapper$$anonfun$startUserClass$2.apply(EGOClusterDriverWrapper.scala:404) at scala.collection.IndexedSeqOptimized$class.foreach(IndexedSeqOptimized.scala:33) at scala.collection.mutable.ArrayOps$ofRef.foreach(ArrayOps.scala:186) at org.apache.spark.deploy.ego.EGOClusterDriverWrapper.startUserClass(EGOClusterDriverWrapper.scala:404) at org.apache.spark.deploy.ego.EGOClusterDriverWrapper.runDriver(EGOClusterDriverWrapper.scala:295) at org.apache.spark.deploy.ego.EGOClusterDriverWrapper.run(EGOClusterDriverWrapper.scala:218) at org.apache.spark.deploy.ego.EGOClusterDriverWrapper$$anonfun$receive$1$$anon$1.run(EGOClusterDriverWrapper.scala:144)

Answer 1

It looks like you have pasted the output from the spark submit script rather than the content of the log files?

You can access the log files with:

curl \
   -X GET \
   -u <tenant_id>:<tenant_secret> \
   -H 'X-Spark-service-instance-id: <instance_id>' \
   https://169.54.219.20/tenant/data/workdir/<submission-id>/stdout

I suggest opening a new question with the error message in the log files if you need more help.

For the second part of your question, notebooks for spark are available through Data Science Experience. You can sign up for Data Science Experience here: https://datascience.ibm.com

Answer 2

I too had faced this error and the problem was that I had not provided a path for the jar in the spark-submit. So when it uploaded the jar there was an error like this in the log:

local2server result: {"content_error":"HTTP request body is empty."}

Adding a path to the jar in the spark-submit helped to resolve my issue, try something like this:

./target/scala-2.10/hellospark_2.10-1.0.jar