hive-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bennie Schut <bsc...@ebuddy.com>
Subject RE: Re:RE: Problem with Hive JDBC server
Date Thu, 07 Feb 2013 09:10:54 GMT
What jdbc driver are you using? Also compiled from trunk? I ask because I remember a jira a
while back where the jdbc driver didn’t let the server know the connection should be closed
().
If that’s the case updating the jdbc driver could work. However that might be a bit of a
long shot.

From: Gabor Makrai [mailto:makrai.list@gmail.com]
Sent: Wednesday, February 06, 2013 12:45 PM
To: 王锋; Bennie Schut
Cc: user@hive.apache.org
Subject: Re: Re:RE: Problem with Hive JDBC server

Hi guys,

Bad news for me. I checked out and compiled the Hive trunk and got the same problem.
I attached to output of command lsof before and after my test program with 100 "SHOW TABLES"
iterations. Is there any explanation why my JDBC server process doesn't release those files?

Thanks,
Gabor

On Tue, Feb 5, 2013 at 6:20 AM, 王锋 <wfeng1982@163.com<mailto:wfeng1982@163.com>>
wrote:


I got it. pls see  https://issues.apache.org/jira/browse/THRIFT-1205

I upgrade the thrift to libthrift-0.9.0.

thanks



At 2013-02-05 13:06:05,"王锋" <wfeng1982@163.com<mailto:wfeng1982@163.com>>
wrote:

When I was using hiveserver ,the exception was thrown:

2060198 Hive history file=/tmp/hdfs/hive_job_log_hdfs_201302010032_1918750748.txt
2060199 Exception in thread "pool-1-thread-95" java.lang.OutOfMemoryError: Java heap space
2060200     at org.apache.thrift.protocol.TBinaryProtocol.readStringBody(TBinaryProtocol.java:353)
2060201     at org.apache.thrift.protocol.TBinaryProtocol.readMessageBegin(TBinaryProtocol.java:215)
2060202     at org.apache.hadoop.hive.service.ThriftHive$Processor.process(ThriftHive.java:730)
2060203     at org.apache.thrift.server.TThreadPoolServer$WorkerProcess.run(TThreadPoolServer.java:253)
2060204     at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1110)
2060205     at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:603)
2060206     at java.lang.Thread.run(Thread.java:722)

I using Hive-0.7.1-cdh3u1 with thrift-0.5.0.jar and thrift-fb303-0.5.0.jar.
how can it be fixed? how about hive-0.7-1 using thrift -0.9.0?  thanks.



At 2013-02-04 19:19:16,"Bennie Schut" <bschut@ebuddy.com<mailto:bschut@ebuddy.com>>
wrote:
Looking at the versions you might be hitting https://issues.apache.org/jira/browse/HIVE-3481
which is fixed in 0.10

On my dev machine the test runs with success :Running time: 298.952409914
This includes this patch so it’s worth looking at.

From: Gabor Makrai [mailto:makrai.list@gmail.com<mailto:makrai.list@gmail.com>]
Sent: Monday, February 04, 2013 11:58 AM
To: user@hive.apache.org<mailto:user@hive.apache.org>
Subject: Re: Problem with Hive JDBC server

Yes, of course! I attached the code!

On Mon, Feb 4, 2013 at 11:57 AM, Gabor Makrai <makrai.list@gmail.com<mailto:makrai.list@gmail.com>>
wrote:
Yes, of course! :) I attached the code!

On Mon, Feb 4, 2013 at 11:53 AM, Bennie Schut <bschut@ebuddy.com<mailto:bschut@ebuddy.com>>
wrote:
Since it’s small can you post the code?

From: Gabor Makrai [mailto:makrai.list@gmail.com<mailto:makrai.list@gmail.com>]
Sent: Monday, February 04, 2013 11:45 AM
To: user@hive.apache.org<mailto:user@hive.apache.org>
Subject: Problem with Hive JDBC server

Hi guys,

I'm writing you because I experienced a very strange problem which probably affects all Hive
distribution.
I made a small "only main function" Java program where I'm only connecting to my Hive JDBC,
and getting the list of the database tables (LIST TABLES) and closing the ResultSet, the Statement
and the Connection and doing this a 1000 times. The problem is that the running Hive JDBC
server does not release files and with time it will throw Exception because, it will get "Too
many open files" IOException from the JVM.

I tested with Hive 0.9, 0.8.1, and the patched Hive 0.9 installed in CDH4.1.1.

If it is a know issue, than could you tell me the solution for it? If it is not, than I can
create a new ticket in Jira, and with a little help, I probably can fix the problem and contribute
the solution for it.

Thanks,
Gabor





Mime
View raw message