Hi community, I have noticed that when a wrong CSV record is ingested and deserialized at CsvRowDeserializationSchema, when ignoreParseErrors is set to false, then an IOException is thrown, which is expected, and in earlier Flink versions, the Flink YARN app would terminated with FINISHED state and FAILED final status for per-job mode Flink jobs without specifically setting a RestartStrategy.
However, to my surprise, in Flink 1.10, the YARN app switched to SUCCEEDED final status afterwards, which is weird since that this Flink job termination was caused by abnormal data input, instead of normal end of source stream. Here I am confused because normally SUCCEEDED final status would give users implications that the job has been **successfully** ended, however this is definitely not the case. So I just wonder if this is the designed behavior of Flink 1.10 on data deserialization errors. Thank you for reading : ) Sincerely, Weike
