perfSONAR User Q&A and Other Discussion

[Jason Zurawski <>]

All;

It is possible for services in perfSONAR-PS to share a port; the 
configuration file for the services can be constructed as such:

<port 9999>

    <endpoint /endpoint1>
        <gls>
        </gls>
    </endpoint>

    <endpoint /endpoint2>
        <perfsonarbuoy>
        </perfsonarbuoy>
    </endpoint>


    <endpoint /endpoint3>
        <snmp>
        </snmp>
    </endpoint>

</port>

There are some caveats to this: the single daemon.pl instance that is 
running the configuration file must know where all of the libraries are 
located.  E.g. if you are thinking about doing this on the Performance 
Toolkit it may be problematic since the services have libraries that are 
split throughout the /usr/local directory.

Thanks;

-jason


> In my opinion, Its not possible, as long as each separate service has
> its own daemon.pl script.
>
> However, if there was just one daemon.pl script that could listen on a
> port, and based on the type of request coming in, could determine, the
> handler to use, and forward it accordingly, this is very much
> possible.
>
>
> Regards,
>
> Fahad Ahmed Satti
> Research Associate
> SLAC
>
>
>
> On Mon, Jan 11, 2010 at 11:38 AM, David Richardson
> <>
>  wrote:
> > On Mon, 11 Jan 2010, Zafar Gilani wrote:
> >
> > > Hey fellow users,
> > >
> > > I was just wondering whether I can shift multiple services to listen on
> > > one port
> > > using one IP address (say for example 4 services mentioned below listen on
> > > xyz.xyz.xyz.xyz:pp):
> > >
> > >
> > > However since it is an important requirement of SLAC's security policy to
> > > keep least
> > > amount of ports in open state, I wanted to ask whether what we are trying
> > > to do is
> > > possible at all. What kind of changes would be required in order to
> > > achieve this? And
> > > whether "single port" approach is recommended or not (and any reasons for
> > > particular
> > > recommendations).
> >
> > It is not possible.
> >
> > A port can only be held by one process at a time. This is an inherent fact,
> > and there is no way to change it.
> >
> > DR
> >
> > --
> > David Richardson 
> > <>
> > "There are two kinds of statistics:
> > the kind you look up and the kind you make up."
> > -- Archie Goodwin, Death of a Doxy