>>
>> Do you have any definitive reasons to believe so, or is it just a (wild)
>> guess based on stack trace?
>
> As I used "likely" it is obviously not definitive, but looking at the
> stack trace, it's in AMX, knowing there is little testing with the JMX
> connector
Sorry. I don't understand this statement.
JMX/HTTP Connector is tested heavily for the portion of it that's
used by asadmin. Note that each asadmin operation invokes JMX/HTTP
Connector implementation.
> and it's a communication error, it is a good start but Lloyd
> needs to chip in anyhow.
>
>>
>> As an aside, does JSR-88 talk about thread-safety of the deployment
>> operations?
>
> why should it ? after all Java is a multithreaded language and unless
> there is explicit provision for things to not be multithreaded, it is
> considered as being capable of handling something as drastically
> complicated as deploying two applications simultaneously.
That does not mean that given implementation is inherently thread-safe.
It does not matter if it is done in Java or not.
Anyway, Lloyd/I will get in touch with Vince to understand the problem
better. Vince, can you please file a bug or refer us to an NB bug if it's
already there?