To whom it may concern I set the parameter "hive.exec.mode.local.auto" to true. Then I used JDBC mode to submit sql to hiveserver2 in a multithreaded manner. Sometimes exceptions are thrown when several tasks run in parallel in the local mode.
"Invalid input path xxx" But this path is not the input path of the current SQL, it is another SQL input path. So I suspect that one of the SQL covers some information about another SQL. So I looked up the source code and found a parameter, and when I set the "iocontext.input.name" to the unique value, I didn't get it wrong. I don't know if this is a bug. I modified the ExecDriver.execute method to add the following line. Random rand = new Random(); job.set(Utilities.INPUT_NAME, "mr_"+rand.nextInt(100000)); Thank you for reading this email. This question has been bothering me for a long time. Your early reply will be appreciated. Thanks &Regards