Royston:
I guess you have seen HBASE-5204. In particular:
>> when a 0.92 server fails to deserialize a 0.90-style RPC, it attempts to
allocate a large buffer because it doesn't read fields of 0.90-style RPCs
properly.

Were your client code compiled with the same version of HBase as what was
running on your cluster ?

Thanks

On Fri, Jan 20, 2012 at 9:20 AM, Royston Sellman <
[email protected]> wrote:

> Trying to run my code (a test of Aggregation Protocol and an MR HBase table
> loader) on latest build of 0.92.0 (r1232715) I get an 'old server' warning
> (I've seen this before and it's always been non-fatal) then an out of
> memory
> exception then job hangs:
>
>
>
> [sshexec] 12/01/20 16:56:48 WARN zookeeper.ClientCnxnSocket: Connected to
> an
> old server; r-o mode will be unavailable
>
>  [sshexec] 12/01/20 16:56:48 INFO zookeeper.ClientCnxn: Session
> establishment complete on server namenode/10.0.0.235:2181, sessionid =
> 0x34cda4e5d000e5, negotiated timeout = 40000
>
>  [sshexec] 12/01/20 16:56:49 WARN ipc.HBaseClient: Unexpected exception
> receiving call responses
>
>  [sshexec] java.lang.OutOfMemoryError: Java heap space
>
>  [sshexec]       at java.lang.reflect.Array.newArray(Native Method)
>
>  [sshexec]       at java.lang.reflect.Array.newInstance(Array.java:52)
>
>  [sshexec]       at
>
> org.apache.hadoop.hbase.io.HbaseObjectWritable.readObject(HbaseObjectWritabl
> e.java:542)
>
>  [sshexec]       at
>
> org.apache.hadoop.hbase.io.HbaseObjectWritable.readFields(HbaseObjectWritabl
> e.java:289)
>
>  [sshexec]       at
>
> org.apache.hadoop.hbase.ipc.HBaseClient$Connection.receiveResponse(HBaseClie
> nt.java:593)
>
>  [sshexec]       at
>
> org.apache.hadoop.hbase.ipc.HBaseClient$Connection.run(HBaseClient.java:505)
>
>
>
> Hbase shell seems to work (I can list and scan my tables).
>
>
>
> If I svn roll back to 12 Jan 0.92 and rebuild my code works.
>
>
>
> Tried setting export HBASE_HEAPSIZE=1500 but got same error.
>
>
>
> Nothing significant in logs.
>
>
>
> [Note to Ted Yu: I need to fix this so I can carry on testing on
> Aggregation
> Protocol]
>
>
>
> Best Regards,
>
> Royston
>
>

Reply via email to