<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
</head>
<body bgcolor="#ffffff" text="#000000">
Hi,<br>
<br>
I was wondering which ports were used by SPREAD and after looking for
the
information in the documentation, I found nothing.<br>
I checked the source code and found in the file network.c that SPREAD
is using port p
(connection in TCP and for data exchange in UDP, i.e. broadcast
channel) defined in spread.conf,
and the port p+1 (UDP channel, i.e. token channel) :<br>
<br>
<tt> Bcast_channel[Num_bcast_channels++] = DL_init_channel(
RECV_CHANNEL, My.port, Bcast_address, interface_addr );<br>
Token_channel[Num_token_channels++] = DL_init_channel(
RECV_CHANNEL, My.port+1, 0, interface_addr );<br>
Send_channel = DL_init_channel( SEND_CHANNEL, My.port+2, 0,
My.id );<br>
</tt><br>
I was surprised to find that port p+2 seems to be used, but after
checking in function <tt>DL_init_channel, </tt>it appears that the
port number is not used for a channel of type SEND_CHANNEL (it could be
nice to replace <tt>My.port+2 </tt>by <tt>dummy_port</tt> for
exmaple)<br>
There was also a presumption that port p+2 was used for monitoring, but
when I checked the monitoring code, I figure out that the port used for
monitoring was provided to the spread deamon by spmonitor (default port
is 6543).<br>
<br>
Can anyone confirm my analyse ?<br>
<br>
On the poject I am working on, our problem is that SPREAD uses
successive ports. We are thinking about an interesting evolution that
would be to have the two ports as parameters in the <tt>spread.conf</tt>
file rather than a computation hard coded. For example :<br>
<tt><br>
Spread_Segment 10.10.3.255:9901:5601 {<br>
...<br>
</tt><br>
Regards,<br>
<div class="moz-signature">-- <br>
<b>Frédéric Maitre</b><font size="1">
</font></div>
</body>
</html>