Re: RMI problem

[EJP <es...@verismartsoftware.com>]

Jukka Zitting wrote:
> 
> By default the RMI server binds itself just to a single IP address,
> 
That's not correct.

By default it binds itself to 0.0.0.0, i.e. INADDR_ANY, just like any other
listening socket does, unless you tell it otherwise by providing an
RMIServerSocketFactory that uses an explicit bind-address.

The issue here is not the bind address but the hostname embedded in the
stub, which is controlled by java.rmi.server.hostname and which has to be
set in the presence of misconfigured DNS or /etc/hosts files or sometimes in
multi-homed hosts. As discussed in the Sun RMI FAQ item A.1.

NB java.rmi.server.hostname has no effect on the bind-address.

EJP

-- 
View this message in context: http://www.nabble.com/RMI-problem-tp15135858p26054418.html
Sent from the Jackrabbit - Users mailing list archive at Nabble.com.

Re: RMI problem

[Jukka Zitting <ju...@gmail.com>]

Hi,

On Mon, Oct 26, 2009 at 5:04 AM, EJP <es...@verismartsoftware.com> wrote:
> Jukka Zitting wrote:
>> By default the RMI server binds itself just to a single IP address,
>>
> That's not correct.
>
> By default it binds itself to 0.0.0.0, i.e. INADDR_ANY, just like any other
> listening socket does, unless you tell it otherwise by providing an
> RMIServerSocketFactory that uses an explicit bind-address.
>
> The issue here is not the bind address but the hostname embedded in the
> stub, which is controlled by java.rmi.server.hostname and which has to be
> set in the presence of misconfigured DNS or /etc/hosts files or sometimes in
> multi-homed hosts. As discussed in the Sun RMI FAQ item A.1.

You're right, thanks for the clarification!

BR,

Jukka Zitting