Merge pull request #57 from WeDataSphere/dev-0.2.3-bugfix

Fix the problem of transformer component in Streamis

Author: jefftlin

streamis-jobmanager/streamis-job-manager/streamis-job-manager-service/src/main/scala/com/webank/wedatasphere/streamis/jobmanager/manager/transform/impl/FlinkJarStreamisJobContentTransform.scala

@@ -27,6 +27,7 @@ import com.webank.wedatasphere.streamis.jobmanager.manager.utils.JobUtils
 
 import scala.collection.JavaConverters._
 import scala.collection.mutable
+import scala.collection.mutable.ArrayBuffer
 
 /**
   * Created by enjoyyin on 2021/9/23.
@@ -50,20 +51,24 @@ class FlinkJarStreamisStartupParamsTransform extends Transform {
       startupMap.put("flink.app.main.class.jar", transformJobContent.getMainClassJar.getFileName)
       startupMap.put("flink.app.main.class.jar.bml.json",
         JsonUtils.jackson.writeValueAsString(getStreamisFileContent(transformJobContent.getMainClassJar)))
-      val classpathFiles = if(transformJobContent.getDependencyJars != null && transformJobContent.getResources != null) {
-        startupMap.put("flink.app.user.class.path", transformJobContent.getDependencyJars.asScala.map(_.getFileName).mkString(","))
-        transformJobContent.getDependencyJars.asScala ++ transformJobContent.getResources.asScala
-      } else if(transformJobContent.getDependencyJars != null) {
-        startupMap.put("flink.app.user.class.path", transformJobContent.getDependencyJars.asScala.map(_.getFileName).mkString(","))
-        transformJobContent.getDependencyJars.asScala
-      } else if(transformJobContent.getResources != null) {
-        startupMap.put("flink.yarn.ship-directories", transformJobContent.getResources.asScala.map(_.getFileName).mkString(","))
-        transformJobContent.getResources.asScala
+
+      /**
+       *  Notice : "flink.app.user.class.path" equals to PipelineOptions.CLASSPATHS in Flink
+       *  paths must specify a protocol (e.g. file://) and be accessible on all nodes
+       *  so we use "flink.yarn.ship-directories" instead
+        */
+      var classPathFiles = Option(transformJobContent.getDependencyJars) match {
+        case Some(list) => list.asScala
+        case _ => mutable.Buffer[StreamisFile]()
+      }
+      Option(transformJobContent.getResources) match {
+        case Some(list) => classPathFiles = classPathFiles ++ list.asScala
+        case _ => // Do nothing
       }
-      else mutable.Buffer[StreamisFile]()
-      if(classpathFiles.nonEmpty)
+      startupMap.put("flink.yarn.ship-directories", classPathFiles.map(_.getFileName).mkString(","))
+      if(classPathFiles.nonEmpty)
         startupMap.put("flink.app.user.class.path.bml.json",
-          JsonUtils.jackson.writeValueAsString(classpathFiles.map(getStreamisFileContent).asJava))
+          JsonUtils.jackson.writeValueAsString(classPathFiles.map(getStreamisFileContent).asJava))
       if(transformJobContent.getHdfsJars != null)
         startupMap.put("flink.user.lib.path", transformJobContent.getHdfsJars.asScala.mkString(","))
       val params = if(job.getParams == null) new util.HashMap[String, Any] else job.getParams