Class RemoteParForUtils

java.lang.Object
org.apache.sysds.runtime.controlprogram.parfor.RemoteParForUtils

public class RemoteParForUtils extends Object
Common functionalities for parfor workers in MR jobs. Used by worker wrappers in mappers (base RemoteParFor) and reducers (fused data partitioning and parfor)
  • Constructor Details

    • RemoteParForUtils

      public RemoteParForUtils()
  • Method Details

    • exportResultVariables

      public static ArrayList<String> exportResultVariables(long workerID, LocalVariableMap vars, ArrayList<ParForStatementBlock.ResultVar> resultVars) throws IOException
      For remote Spark parfor workers. This is a simplified version compared to MR.
      Parameters:
      workerID - worker id
      vars - local variable map
      resultVars - list of result variables
      Returns:
      list of result variables
      Throws:
      IOException - if IOException occurs
    • exportLineageItems

      public static void exportLineageItems(long workerID, LocalVariableMap vars, ArrayList<ParForStatementBlock.ResultVar> resultVars, Lineage lineage) throws IOException
      Export lineage for remote Spark parfor workers.
      Parameters:
      workerID - worker id
      vars - local variable map
      resultVars - list of result variables
      lineage - lineage object
      Throws:
      IOException - if IOException occurs
    • cleanupWorkingDirectories

      public static void cleanupWorkingDirectories()
      Cleanup all temporary files created by this SystemDS process.
    • cleanupWorkingDirectoriesOnShutdown

      public static void cleanupWorkingDirectoriesOnShutdown()
      Cleanup all temporary files created by this SystemDS process, on shutdown via exit or interrupt.
    • getLineages

      public static Lineage[] getLineages(LocalVariableMap[] results)
    • getResults

      public static LocalVariableMap[] getResults(List<scala.Tuple2<Long,String>> out, org.apache.commons.logging.Log LOG)
    • setupBufferPool

      public static void setupBufferPool(long workerID, boolean isLocal) throws IOException
      Init and register-cleanup of buffer pool
      Parameters:
      workerID - worker id
      isLocal - in local spark mode (single JVM)
      Throws:
      IOException - exception