[GitHub] [flink] shuiqiangchen commented on a change in pull request #13322: [FLINK-17480][kubernetes] Support running PyFlink on Kubernetes.

[GitBox <gi...@apache.org>]

shuiqiangchen commented on a change in pull request #13322:
URL: https://github.com/apache/flink/pull/13322#discussion_r491760113



##########
File path: flink-clients/src/main/java/org/apache/flink/client/cli/ProgramOptionsUtils.java
##########
@@ -93,4 +98,40 @@ public static ProgramOptions createPythonProgramOptions(CommandLine line) throws
 					"or not working as expected.", e);
 		}
 	}
+
+	public static ProgramOptions createPythonApplicationProgramOptions(CommandLine line) throws CliArgsException,
+		NoSuchFieldException, IllegalAccessException {
+		ProgramOptions pythonProgramOptions = createPythonProgramOptions(line);
+		Field jarFilePath = pythonProgramOptions.getClass().getSuperclass().getDeclaredField("jarFilePath");
+		jarFilePath.setAccessible(true);
+		// This is the python jar path in client, which is invalid at runtime and it will be replaced with the actual
+		// path when retrieving the packaged program in the job manager container.
+		String pythonJarPath = "local:///opt/flink/opt/" + FilenameUtils.getName(PackagedProgramUtils.getPythonJar()

Review comment:
       Here we have two options for PyFlink job:
   1. Make a judgement that if it is PyFlink job, skip the jarFile path validation;
   2. Set a dummy jar file path in client side and replace it with the actual path in JM.
   It seems that the former is better.




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org