You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@river.apache.org by Sergio Aguilera Cazorla <sa...@gmail.com> on 2011/09/22 16:57:20 UTC

Reggie's visibility in discovery process

Hello,

I am a telecommunication enginnering student and I'm working on my final
degree thesis using Apache River technology. My goal is to integrate the
distributed capabilities that River offers with a previously developed
software written in the Java programming language.

I've been working for some months using reggie as the Lookup Server. I've
always tried my programs using a single machine. I've been careful in the
sense that I've executed client, server and reggie in different locations,
different JVM's, etc. However, problems arise when we try the software
developed so far in 3 different machines: we are unable to get a response
from reggie. We are executing reggie with the default configuration provided
with the "hello" example.

As everything works fine in a single machine with the proper cautions to
simulate a distributed system, we suspect that the problem is related to
network configuration, firewalls and switches. We perform multicast
discovery in order to contact reggie (yes, unicast discovery may work, but
the software we are developing would be pointless if we needed a specific
server URL).

May there be some problems with multicast packets? Could someone post a list
of possible reasons that could make the visibility of reggie crash if it
works fine in a single machine?

Next days we'll perform some more testing and hopefully I'll be able to
provide you with more data.

Thanks in advance,

-- 
*Sergio Aguilera*

Re: Reggie's visibility in discovery process

Posted by Zsolt Kúti <la...@gmail.com>.

Hi,

LookupDiscovery uses multicast, so your suspect looks the likely source of
the problem. Check the network setup between/within machines.

Zsolt

On Thu, Sep 22, 2011 at 5:30 PM, Sergio Aguilera Cazorla
<sa...@gmail.com>wrote:

> Hello,
>
> The only specific data that I can provide at this moment is the piece of
> code that is used to perform discovery from the server-side:
>
> DiscoveryManagement dlm = new LookupDiscovery(LookupDiscovery.ALL_GROUPS);
>
> ServiceItem item = new ServiceItem(//...doesn't matter);
>
>        try {
>            Thread.sleep(2000);    //To give time to the Lookup Server to
> respond
>        } catch (InterruptedException e) { e.printStackTrace(); }
>
>        ServiceRegistrar[] srList = dlm.getRegistrars();
>
> //perform the registration with the found Lookups
>
>
> I relied in this code from the first day, and I constructed my software
> making for sure that it worked. Problems only appear when I move the Lookup
> execution to a different machine than the one that is executing the server.
>
> More details: we are in a production environment (a company's LAN), so
> maybe
> we are missing some special security configurations that we are not used to
> (some multicast packet filtering maybe?). Also, if I increase the waiting
> time you can see in the code from 2 seconds to 10, 20, 30... it still
> doesn't work.
>
> I have a list of tests to perform tomorrow, if I can meet with 2 mates to
> use the 3-machine configuration (server-lookup-client). I'll post here our
> conclusions, if we reach some, but in the meanwhile if you have any idea of
> the source of the problem, it may help us in tomorrow's test.
>
> Regards,
>
> --
> *Sergio Aguilera*
>