- Fixed bug in PBS template that prevented correct

  error message when deployment failed.
- Fixed bug in deployment script that prevented
  correct error message when exception was thrown

spark_deploy.py

@@ -131,7 +131,7 @@ def execCommand( a_command ):
     retcode = proc.poll()
 
     if retcode:
-        raise subprocess.CalledProcessError( retcode, command, output = output[0] )
+        raise subprocess.CalledProcessError( retcode, a_command, output = output[0] )
 
     return output
 

spark_setup.py

@@ -84,7 +84,7 @@ def generateScript():
     outf.write( "    sleep 1\n" )
     outf.write( "    let timeout=timeout-1\n" )
     outf.write( "    if [ $timeout -eq 0 ]; then\n" )
-    outf.write( "       >&2 Spark deployment failed!\n" )
+    outf.write( "       echo >&2 Spark deployment failed. Inspect per-node logfiles in deployment directory for diagnostics.\n" )
     outf.write( "       exit 1\n" )
     outf.write( "    fi\n" )
     outf.write( "done\n\n" )