This really sounds like you have bits of different hadoop versions floating
around.  This could be in the cluster itself rather than on your dev
machine.  Client and server here problem mean map-reduce program and
datanode or namenode.

It is even conceivably a CDH bug that isn't tickled by other code.  That
seems really unlikely, though, both because I don't understand how we would
tickle a piece of code nobody else has and because the RPC code is pretty
centralized.

On Sat, May 21, 2011 at 10:43 AM, Jeff Eastman
<[email protected]>wrote:

> Version Mismatch between client and server... command aborted.
>
> It seems to be thrown by the DFSAdmin class run() method when an
> RPC.VersionMismatch exception is thrown. I verified I have the same CDH3
> client and server versions so I'm baffled.  I did upgrade my client CDH3
> from the February 21 beta to the March 25 production bits, but I still get
> the error. All of the many other jobs I've submitted to that cluster from my
> client VM work without issue. Is 20 newsgroups doing anything unusual?
>
>
> On 5/21/11 3:46 AM, Grant Ingersoll wrote:
>
>> Jeff, what's your other exception?
>>
>>
>

Reply via email to