Patrick Liu created SPARK-3544:
----------------------------------
Summary: SparkSQL thriftServer cannot release locks correctly in
Zookeeper
Key: SPARK-3544
URL: https://issues.apache.org/jira/browse/SPARK-3544
Project: Spark
Issue Type: Bug
Components: Spark Core
Affects Versions: 1.1.0
Reporter: Patrick Liu
Priority: Critical
Bug description:
The thriftServer cannot release locks correctly in Zookeeper.
Once a thriftServer is started, the 1st SQL submited by Beeline or JDBC
requiring locks can be successfully computed.
However, the 2rd SQL requiring locks will be blocked. And the thriftServer log
shows: INFO Driver: <PERFLOG method=acquireReadWriteLocks>.
2 Tests to illustrate the problem:
Test 1:
(0) Start thriftServer & use beeline to connect to it.
(1) Switch database(require locks);(OK)
(2) Drop table(require locks) “drop table src"; (Blocked)
Test 2:
(0) Start thriftServer & use beeline to connect to it.
(1) Drop table(require locks) "drop table src"; (OK)
(2) Drop another table(require locks) "drop table src2;"(Blocked)
Basic Information:
Spark 1.1.0
Hadoop 2.0.0-cdh4.6.0
Compile Command:
export MAVEN_OPTS="-Xmx2g -XX:MaxPermSize=512M -XX:ReservedCodeCacheSize=512m"
mvn -Dhadoop.version=2.0.0-cdh4.6.0 -Phive -Pspark-ganglia-lgpl -DskipTests
package
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
