[HIVE-13626] Unable to cancel certain queries [Spark Branch]

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: Spark

Environment

CDH5.7 single node docker

• Assignee: Unassigned
• Reporter: Zsolt Tóth
• Created: 27/Apr/16 13:08
• Updated: 27/Apr/16 13:08

Description

To reproduce:
1. Submit a query to Hive through JDBC
2. Cancel (almost) immediately

For the query SELECT DISTINCT a FROM test the Hive log says: "java.lang.Exception: Failed to submit Spark work, please retry later" ... "Caused by: java.io.IOException: java.lang.InterruptedException". After I call close() on the statement and the connection, the application changes to FINISHED/SUCCEEDED state (as expected).

However, for the query SELECT a FROM test order by a Hive log says: "org.apache.hadoop.hive.ql.metadata.HiveException: Failed to create spark client." ... "Caused by: java.lang.RuntimeException: java.lang.InterruptedException". Calling close() seems to have no effect, the yarn application is stuck in RUNNING state.

If I cancel the second query later (e.g. after 10 secs), the application closes with FINISHED/SUCCEEDED (as expected).

Code snippet:

Bar.java

final Statement st = connection.createStatement();
new Thread() {
	@Override
	public void run() {
		sleep(100);
		st.cancel();
	}
}.start();

ResultSet r = null;
try {
	// r = st.executeQuery("SELECT DISTINCT a FROM test");
	r = st.executeQuery("SELECT a FROM test order by a");
} catch (SQLException e) {
	close(r);
	close(st);
	close(connection);
}