0 Session Length, radrelay or Freeradius causing issues
Hello List, I have asked this before (http://readlist.com/lists/lists.freeradius.org/freeradius-users/1/7788.html) and Alan DeKok did say that I should place those accounting packets in a different Acct-Type, but that was when I did not wish to log the information but I now need to log the information and replicate to all radius servers. I need to know why the radrelay application stops replacation when it comes across a 0 session length packet. This morning I found a 40Mb backlog file for each server and realised that it was the 0 session-length packet again. I know FreeRadius does complain about it but it still logs this accounting packet into the sql database. So is it because FreeRadius indicates that an error (0 Session Length?). If so could I not just somehow tell FreeRadius to stop complaining and just accept the stupid packet. (As I am passing configurable triggers via 0 session length radius packets). -- Kind Regards Etienne Pretorius
Etienne Pretorius wrote:
I need to know why the radrelay application stops replacation when it comes across a 0 session length packet.
Because the RADIUS server says that the accounting packet was not processed, so radrelay tries to send it again. This is what a NAS does when it does not receive a reply to an accounting request.
This morning I found a 40Mb backlog file for each server and realised that it was the 0 session-length packet again.
I know FreeRadius does complain about it but it still logs this accounting packet into the sql database. So is it because FreeRadius indicates that an error (0 Session Length?). If so could I not just somehow tell FreeRadius to stop complaining and just accept the stupid packet. (As I am passing configurable triggers via 0 session length radius packets).
You need to tell the RADIUS server that it should respond to the accounting request with 0 session length. Grab CVS head, and read raddb/sites-available/copy-acct-to-home-server It explains this. Also, in 2.0, "radrelay" is part of the server, and works much better than in 1.1.x. Alan DeKok.
Alan DeKok wrote:
Etienne Pretorius wrote:
I need to know why the radrelay application stops replacation when it comes across a 0 session length packet.
Because the RADIUS server says that the accounting packet was not processed, so radrelay tries to send it again. This is what a NAS does when it does not receive a reply to an accounting request.
This morning I found a 40Mb backlog file for each server and realised that it was the 0 session-length packet again.
I know FreeRadius does complain about it but it still logs this accounting packet into the sql database. So is it because FreeRadius indicates that an error (0 Session Length?). If so could I not just somehow tell FreeRadius to stop complaining and just accept the stupid packet. (As I am passing configurable triggers via 0 session length radius packets).
You need to tell the RADIUS server that it should respond to the accounting request with 0 session length.
Grab CVS head, and read raddb/sites-available/copy-acct-to-home-server
It explains this. Also, in 2.0, "radrelay" is part of the server, and works much better than in 1.1.x.
Alan DeKok. - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
Thank you. I'll take a look. Kind Regards Etienne Pretorius
Hi,
Hello List,
I have asked this before (http://readlist.com/lists/lists.freeradius.org/freeradius-users/1/7788.html) and Alan DeKok did say that I should place those accounting packets in a different Acct-Type, but that was when I did not wish to log the information but I now need to log the information and replicate to all radius servers.
I need to know why the radrelay application stops replacation when it comes across a 0 session length packet. This morning I found a 40Mb backlog file for each server and realised that it was the 0 session-length packet again.
I know FreeRadius does complain about it but it still logs this accounting packet into the sql database. So is it because FreeRadius indicates that an error (0 Session Length?). If so could I not just somehow tell FreeRadius to stop complaining and just accept the stupid packet. (As I am passing configurable triggers via 0 session length radius packets).
I have similar issues (had similar issues) and my detail config now only fires stuff into the detail log for 'out of band' SQL processing if that Acct-Session-Time does not = 0. however, I believe that the latest CVS version (so this will be in 2.0.5) now handles session time of 0 alan
participants (3)
-
A.L.M.Buxey＠lboro.ac.uk -
Alan DeKok -
Etienne Pretorius