Re: Compiling Torque with non-default ports

Hi Gareth,

Thanks for the response. But I found out that it was as simple as my  
LD_LIBRARY_PATH having the original Torque library ahead of the new one.

Once I changed that, I could see the different Torque installations  
just fine.

Thanks again,
Prakash

On Aug 23, 2007, at 8:39 PM, <Gareth.Williams@xxxxxxxx> wrote:

> Hi Prakash,
>
> Torque looks through /etc/services to lookup ports for services by  
> name.
> I have in the past patched pbs_ifl.h to change the service names as  
> well
> as the port numbers, and added the new names/numbers to /etc/ 
> services to
> get two complete torque installations that do not interfere.
>
> Also, I renamed the binaries so I could setup distinct /etc/init.d
> scripts and not get confused about which daemon was which.
>
> Gareth Williams
> Outreach Manager
> CSIRO IM&T - eScience (HPSC)
>
> > -----Original Message-----
> > From: "Prakash Velayutham" <prakash.velayutham@xxxxxxxxx>
> > To: torqueusers@xxxxxxxxxxxxxxxx
> > Sent: 8/10/07 1:56 PM
> > Subject: [torqueusers] Compiling Torque with non-default ports
> >
> > Hello,
> >
> > I recently compiled Torque (for testing) with the port numbers and
> > names changed from 15001, 15002, 15003 and 1504 in the file
> > pbs_ifl.h. After installation, if I start up the server, scheduler
> > and mom, everything is running well and listening on the
> > non-standard
> > ports that I asked.
> >
> > I also have the production Torque running in the same server at the
> > standard ports. Now, if I run the non-standard binary pbsnodes or
> > qstat, strangely it is pulling information off the production server
> > instance. Anyone knows why this is the case? I thought when I modify
> > the port names and numbers in the pbs_ifl.h file, automatically the
> > default port to listen on and connect to for server, scheduler, mom
> > and clients are changed.
> >
> > Thanks and have a great day tomorrow,
> > Prakash