I have been trying Spark 2.4 deployment on k8s and want to establish a secured RPC communication channel between driver and executors. Was using the following configuration parameters as part of spark-submit
spark.authenticate true
spark.authenticate.secret good
spark.network.crypto.enabled true
spark.network.crypto.keyFactoryAlgorithm PBKDF2WithHmacSHA1
spark.network.crypto.saslFallback falseThe driver and executors were not able to communicate on a secured channel and were throwing the following errors.
Exception in thread "main" java.lang.reflect.UndeclaredThrowableException
at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1713)
at org.apache.spark.deploy.SparkHadoopUtil.runAsSparkUser(SparkHadoopUtil.scala:64)
at org.apache.spark.executor.CoarseGrainedExecutorBackend$.run(CoarseGrainedExecutorBackend.scala:188)
at org.apache.spark.executor.CoarseGrainedExecutorBackend$.main(CoarseGrainedExecutorBackend.scala:281)
at org.apache.spark.executor.CoarseGrainedExecutorBackend.main(CoarseGrainedExecutorBackend.scala)
Caused by: org.apache.spark.SparkException: Exception thrown in awaitResult:
at org.apache.spark.util.ThreadUtils$.awaitResult(ThreadUtils.scala:226)
at org.apache.spark.rpc.RpcTimeout.awaitResult(RpcTimeout.scala:75)
at org.apache.spark.rpc.RpcEnv.setupEndpointRefByURI(RpcEnv.scala:101)
at org.apache.spark.executor.CoarseGrainedExecutorBackend$anonfun$run$1.apply$mcV$sp(CoarseGrainedExecutorBackend.scala:201)
at org.apache.spark.deploy.SparkHadoopUtil$anon$2.run(SparkHadoopUtil.scala:65)
at org.apache.spark.deploy.SparkHadoopUtil$anon$2.run(SparkHadoopUtil.scala:64)
at java.security.AccessController.doPrivileged(Native Method)
at javax.security.auth.Subject.doAs(Subject.java:422)
at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1698)
... 4 more
Caused by: java.lang.RuntimeException: java.lang.IllegalArgumentException: Unknown challenge message.
at org.apache.spark.network.crypto.AuthRpcHandler.receive(AuthRpcHandler.java:109)
at org.apache.spark.network.server.TransportRequestHandler.processRpcRequest(TransportRequestHandler.java:181)
at org.apache.spark.network.server.TransportRequestHandler.handle(TransportRequestHandler.java:103)
at org.apache.spark.network.server.TransportChannelHandler.channelRead(TransportChannelHandler.java:118)
at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:362)Can someone guide me on this?
Disclaimer: I do not have a very deep understanding of spark implementation, so, be careful when using the workaround described below.
AFAIK, spark does not have support for auth/encryption for k8s in 2.4.0 version.
There is a ticket, which is already fixed and likely will be released in a next spark version: https://issues.apache.org/jira/browse/SPARK-26239
The problem is that spark executors try to open connection to a driver, and a configuration will be sent only using this connection. Although, an executor creates the connection with default config AND system properties started with "spark.". For reference, here is the place where executor opens the connection: https://github.com/apache/spark/blob/5fa4384/core/src/main/scala/org/apache/spark/executor/CoarseGrainedExecutorBackend.scala#L201
Theoretically, if you would set spark.executor.extraJavaOptions=-Dspark.authenticate=true -Dspark.network.crypto.enabled=true ..., it should help, although driver checks that there are no spark parameters set in extraJavaOptions.
Although, there is a workaround (a little bit hacky): you can set spark.executorEnv.JAVA_TOOL_OPTIONS=-Dspark.authenticate=true -Dspark.network.crypto.enabled=true .... Spark does not check this parameter, but JVM uses this env variable to add this parameter to properties.
Also, instead of using JAVA_TOOL_OPTIONS to pass secret, I would recommend to use spark.executorEnv._SPARK_AUTH_SECRET=<secret>.