[omniORB] rendezvouser error

[Duncan Grisby]

On Wednesday 22 June, "John Chesshir" wrote:

>     Here is some more info.  I tried putting the traceLevel environment
> variable to 40.  I have attached results from the Red Hat Linux 7.0
> server (server.bad) and the Red Hat Linux 7.2 2.96-128.7.2 server
> (server.good)  The only difference besides the IP address is that the
> bad one notes a failure so spawn a thread and then the cannot create
> rendezvouser message, while the good one notes a successful thread
> startup and a rendezvouser task set up.  
>     Also, I got information on the hardware.  Red Hat 7.0 is running on
> an IBM x390 server, while the other is running on a dell PIII desktop.
>     I am thinking this may have to do with Redhat or the Dell not being
> able to spawn threads, although I do see two processes with my server's
> name on them where there have generally been three on startup ever since
> we went to omniORB.  I always assumed the extra two processes were Linux
> threads, although the trace only shows one thread being (or trying to
> be) created.  Is that other sub-process something else?  Can anyone
> confirm, deny, or give extra thoughts?

The "cannot create a rendezvouser" message is indeed a sign that omniORB
was unable to start a thread. You need to understand why that is. I'd
suggest trying to write a simple program that starts a thread, first
with plain pthread calls, then with omnithread, to see whether that
works.

omniORB definitely works fine on RedHat 7.0 for Intel processors, but I
don't know about the x390.

On RedHat 7 era linux, you see one process in the process listing for
each thread, plus one for a manager thread, so that explains the extra
process you see.

Cheers,

Duncan.

-- 
 -- Duncan Grisby         --
  -- duncan at grisby.org     --
   -- http://www.grisby.org --