Re: jobtracker.jsp reports "GC overhead limit exceeded"

[Hemanth Yamijala <yh...@gmail.com>]

Hi,

> Actually I enabled all level logs. But I didn't realize to check logs in .out files and only looked at .log file and didn't see any error msgs. now I opened the .out file and saw the following logged exception:
>
> Exception in thread "IPC Server handler 5 on 50002" java.lang.OutOfMemoryError: Java heap space
>     at java.util.Arrays.copyOf(Arrays.java:2786)
>     at java.io.ByteArrayOutputStream.toByteArray(ByteArrayOutputStream.java:133)
>     at org.apache.hadoop.ipc.Server.setupResponse(Server.java:1087)
>     at org.apache.hadoop.ipc.Server.access$2400(Server.java:77)
>     at org.apache.hadoop.ipc.Server$Handler.run(Server.java:984)
>
> Exception in thread "IPC Client (47) connection to servername/serverip:50001 from auser" java.lang.OutOfMemoryError: Java heap space
>     at java.nio.ByteBuffer.wrap(ByteBuffer.java:350)
>     at org.apache.hadoop.net.SocketInputStream.read(SocketInputStream.java:128)
>     at java.io.FilterInputStream.read(FilterInputStream.java:116)
>     at org.apache.hadoop.ipc.Client$Connection$PingInputStream.read(Client.java:276)
>     at java.io.BufferedInputStream.fill(BufferedInputStream.java:218)
>     at java.io.BufferedInputStream.read(BufferedInputStream.java:237)
>     at java.io.DataInputStream.readInt(DataInputStream.java:370)
>     at org.apache.hadoop.ipc.Client$Connection.receiveResponse(Client.java:501)
>     at org.apache.hadoop.ipc.Client$Connection.run(Client.java:446)
>
> Exception in thread "LeaseChecker" java.lang.OutOfMemoryError: GC overhead limit exceeded
> Exception in thread "IPC Server handler 4 on 50002" java.lang.OutOfMemoryError: Java heap space
> Exception in thread "IPC Server handler 8 on 50002" java.lang.OutOfMemoryError: GC overhead limit exceeded
> Exception in thread "IPC Server handler 9 on 50002" java.lang.OutOfMemoryError: GC overhead limit exceeded
> Exception in thread "IPC Server handler 2 on 50002" java.lang.OutOfMemoryError: GC overhead limit exceeded
> Exception in thread "IPC Server handler 6 on 50002" java.lang.OutOfMemoryError: GC overhead limit exceeded
> Exception in thread "ResponseProcessor for block blk_-573819335330670501_1268378" java.lang.OutOfMemoryError: GC overhead limit exceeded
> Exception in thread "IPC Server handler 7 on 50002" java.lang.OutOfMemoryError: Java heap space
>
> 50001 and 50002 used for dfs and jobtracker. But it's normal that IPC threads consume such big memory?
>

It may not be IPC threads that are causing the problem, though I
remember at least one IPC buffer related OOM somewhere. A more typical
case is that the daemon is running out of memory and the IPC threads
are just manifesting the error because they're the entry-points into
the application. I think we still need to go down the path that Alex
was suggesting. What is the heap you have given the JobTracker ? What
is number of Jobs / tasks / users ? How many nodes are there in the
cluster ?

>
> --- On Fri, 7/30/10, Alex Loddengaard <al...@cloudera.com> wrote:
>
> From: Alex Loddengaard <al...@cloudera.com>
> Subject: Re: jobtracker.jsp reports "GC overhead limit exceeded"
> To: common-user@hadoop.apache.org
> Date: Friday, July 30, 2010, 2:19 PM
>
> What does "ps" show you?  How much memory is being used by the jobtracker,
> and how large is its heap (loop for HADOOP_HEAPSIZE in hadoop-env.sh)?  Also
> consider turning on GC logging, which will find its way to the jobtracker
> .out log in /var/log/hadoop:
>
> <http://java.sun.com/developer/technicalArticles/Programming/GCPortal/>
>
> Alex
>
> On Fri, Jul 30, 2010 at 3:10 PM, jiang licht <li...@yahoo.com> wrote:
>
>> http://server:50030/jobtracker.jsp generates the following error message:
>>
>> HTTP ERROR: 500
>>
>> GC overhead limit exceeded
>>
>> RequestURI=/jobtracker.jsp
>> Caused by:
>>
>> java.lang.OutOfMemoryError: GC overhead limit exceeded
>>
>> Powered by Jetty://
>>
>> The jobtracker is running below the limit. But "hadoop job -status" seems
>> to halt and does not response ...
>>
>> The last 2 lines of jobtracker logs:
>>
>> 2010-07-30 13:53:18,482 DEBUG org.apache.hadoop.mapred.JobTracker: Got
>> heartbeat from: tracker_host1:localhost.localdomain/127.0.0.1:53914(restarted: false initialContact: false acceptNewTasks: true) with
>> responseId: -31252
>> 2010-07-30 13:55:32,917 DEBUG org.apache.hadoop.mapred.JobTracker: Starting
>> launching task sweep
>>
>> Any thought about this?
>>
>> Thanks!
>> --Michael
>>
>>
>>
>
>
>
>