Github user yanghua commented on a diff in the pull request: https://github.com/apache/flink/pull/5448#discussion_r197736638 --- Diff: flink-clients/src/main/java/org/apache/flink/client/deployment/ClusterSpecification.java --- @@ -67,8 +68,8 @@ public String toString() { public static ClusterSpecification fromConfiguration(Configuration configuration) { int slots = configuration.getInteger(TaskManagerOptions.NUM_TASK_SLOTS, 1); - int jobManagerMemoryMb = configuration.getInteger(JobManagerOptions.JOB_MANAGER_HEAP_MEMORY); - int taskManagerMemoryMb = configuration.getInteger(TaskManagerOptions.TASK_MANAGER_HEAP_MEMORY); + int jobManagerMemoryMb = (int) MemorySize.parse(configuration.getString(JobManagerOptions.JOB_MANAGER_HEAP_MEMORY)).getMebiBytes(); --- End diff -- hi @dawidwys this belongs suggestion 2, "change the return value of getMebiBytes() to int or have a getMebiBytesAsInt() method that uses a MathUtils.checkedDownCast() to avoid unnoticed overflow errors, as Stephan commented" "have a getMebiBytesAsInt()" is another choice, if I picked the first choice and change the `(int) to MathUtils.checkedDownCast()` . Is there necessary to provide the `getMebiBytesAsInt()` method?
---