Re: [openorb-users] OpenORB NamingService: java.io.EOFException

[<lk...@us...>]

On 5/23/07, rosmi.jose wrote:

> Yes, the Naming service process stop on the OS level.

I haven't rechecked it, but as far as I know there is no explicit
System.exit() call in the core ORB classes or the NamingService. Here
are a few questions that might lead towards finding the problem:

* Are you using tns or ins? If ins, which persistence mechanism (file,
db)? Any error messages on startup?
* Under which circumstances does the NamingServer terminate, during
it's startup, during the first client request, or any other specific
situation?
* Have you tried to increase logging output or to attach to the
NamingService with a debugger?
* Are you sure that nothing in your system sends the OS process a
termination signal (or whatever it's called on non-UNIX platforms)?

> CORBA also stopped.

What do you mean by this? CORBA is a not something that starts and
stops (like a OS process). It's a specification, much like EJB. What
exactly stops?

> We are facing this problem System Verification phase.
>
> OpenORB Version - 1.3.1

OpenORB 1.4.0 has been released about 2 years ago. Why are you using a
version that is even older than that? Loads of bugs have been fixed
since 1.3.1. Have you tried upgrading?

> Since corba is going down,

You mean the NamingService process is going down?

> client and server side applications undergo
> shutdown. It causes the related services to go down.

I'm pretty sure that this is either caused by something in your system
monitoring (some monitoring program decides "oh, process XYZ depends
on the NamingService, that went down so I'll kill XYZ as well") or by
badly written client code that doesn't handle SystemExceptions
gracefully.

> Can you explain me the scenario's(general) where in CORBA keep restarting/failing? Why does Corba keep restarting/failing?

Nothing in CORBA / OpenORB does this.

Regards,
Lars