Hi Harshit,

On 19.04.21, 10:48, "Harshit Jain" hjain@arista.com wrote:

I see that we set SO_REUSEADDR to false while binding to a datagram socket whereas we set it to true for stream sockets. Is there a particular reason for this?

Not sure about alle the implications, and if it would event work. But basically if you set SO_REUSEADDR on a DGRAM socket, every socket bound to that port/address will receive a copy of the packet. In your case, using the same port to listen and as a source for the outgoing packets, both threads (serving radsrv() and replyh() ) would get all requests and responses.

In my setup, I have radsecproxy listening on *:2083 for UDP requests and I have a local radius server listening on localhost:3799 (among other external servers). I was looking to set a fixed source port from which radsecproxy sends out requests to the local radius server, so I added a source block to the server config like so server localServer { ... source 127.0.0.2:2083 } but I got an error saying "address already in use". I dug around and found that its because SO_REUSEADDR is not set for datagram sockets. Is there a reason for this and if so, is there any other way to add a fixed source port to a local radius server config block?

Telling by your config snippet, you are already using the master branch, which basically allows to specify a source port/address per server (this will be included in the next 1.9 release). However, because auf the reasoning above, you cannot use the same port you are already listening on; at least for UDP.

Please note that while having a second look at the code, I got the feeling that there might still be some bugs in there.

Regards, Fabian

-- SWITCH Fabian Mauchle, Network Engineer Werdstrasse 2, P.O. Box, 8021 Zurich, Switzerland Phone +41 44 268 15 30, direct +41 44 268 15 39