[
https://issues.apache.org/jira/browse/HIVE-5296?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13769320#comment-13769320
]
Douglas commented on HIVE-5296:
-------------------------------
Hi Kousuke,
Thanks for your interest. Here are the answers to your questions:
1) This is most definitely the Hiverserver2 process. I validated this by
tracking the heap space utilised for the hiveserver2 process over time, as
connections were opened and closed.
2) The queries that were being executed were for the most part "LOAD DATA
INPATH:"
{code}
int returnCode = hc.update("LOAD DATA INPATH \'"+fileName +"\' " +
"OVERWRITE INTO TABLE "+targetTable+"
partition (dt=\'"+cal.getTimeInMillis()+"\')");
logger.info(this.getClass()+ " returned with value
"+returnCode);
{code}
These were a mix of successes, and exceptions. I've yet to validate if the leak
occurs in all instances, or in those cases where the hiveserver throws
Exceptions.
3) I've not had the time to dig into the hiveserver code as yet to find the
offending object, but if I do get the chance, I will certainly post my findings
and a patch.
> Memory leak: OOM Error after multiple open/closed JDBC connections.
> --------------------------------------------------------------------
>
> Key: HIVE-5296
> URL: https://issues.apache.org/jira/browse/HIVE-5296
> Project: Hive
> Issue Type: Bug
> Components: HiveServer2
> Affects Versions: 0.12.0
> Environment: Hive 0.12.0, Hadoop 1.1.2, Debian.
> Reporter: Douglas
> Labels: hiveserver
> Fix For: 0.12.0
>
> Original Estimate: 168h
> Remaining Estimate: 168h
>
> This error seems to relate to https://issues.apache.org/jira/browse/HIVE-3481
> However, on inspection of the related patch and my built version of Hive
> (patch carried forward to 0.12.0), I am still seeing the described behaviour.
> Multiple connections to Hiveserver2, all of which are closed and disposed of
> properly show the Java heap size to grow extremely quickly.
> This issue can be recreated using the following code
> {code}
> import java.sql.DriverManager;
> import java.sql.Connection;
> import java.sql.ResultSet;
> import java.sql.SQLException;
> import java.sql.Statement;
> import java.util.Properties;
> import org.apache.hive.service.cli.HiveSQLException;
> import org.apache.log4j.Logger;
> /*
> * Class which encapsulates the lifecycle of a query or statement.
> * Provides functionality which allows you to create a connection
> */
> public class HiveClient {
>
> Connection con;
> Logger logger;
> private static String driverName = "org.apache.hive.jdbc.HiveDriver";
> private String db;
>
>
> public HiveClient(String db)
> {
> logger = Logger.getLogger(HiveClient.class);
> this.db=db;
>
> try{
> Class.forName(driverName);
> }catch(ClassNotFoundException e){
> logger.info("Can't find Hive driver");
> }
>
> String hiveHost = GlimmerServer.config.getString("hive/host");
> String hivePort = GlimmerServer.config.getString("hive/port");
> String connectionString = "jdbc:hive2://"+hiveHost+":"+hivePort
> +"/default";
> logger.info(String.format("Attempting to connect to
> %s",connectionString));
> try{
> con =
> DriverManager.getConnection(connectionString,"","");
>
> }catch(Exception e){
> logger.error("Problem instantiating the
> connection"+e.getMessage());
> }
> }
>
> public int update(String query)
> {
> Integer res = 0;
> Statement stmt = null;
> try{
> stmt = con.createStatement();
> String switchdb = "USE "+db;
> logger.info(switchdb);
> stmt.executeUpdate(switchdb);
> logger.info(query);
> res = stmt.executeUpdate(query);
> logger.info("Query passed to server");
> stmt.close();
> }catch(HiveSQLException e){
> logger.info(String.format("HiveSQLException thrown,
> this can be valid, " +
> "but check the error: %s from the query
> %s",query,e.toString()));
> }catch(SQLException e){
> logger.error(String.format("Unable to execute query
> SQLException %s. Error: %s",query,e));
> }catch(Exception e){
> logger.error(String.format("Unable to execute query %s.
> Error: %s",query,e));
> }
>
> if(stmt!=null)
> try{
> stmt.close();
> }catch(SQLException e){
> logger.error("Cannot close the statment,
> potentially memory leak "+e);
> }
>
> return res;
> }
>
> public void close()
> {
> if(con!=null){
> try {
> con.close();
> } catch (SQLException e) {
> logger.info("Problem closing connection "+e);
> }
> }
> }
>
>
>
> }
> {code}
> And by creating and closing many HiveClient objects. The heap space used by
> the hiveserver2 runjar process is seen to increase extremely quickly, without
> such space being released.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
