Just for fun, can you post your default radius entry? Also, what shows in accounting? The 0.0.0.0 address or a proper IP? Anything in common on the client side (all the same OS?)? You could also try setting up a local user on the boxes and give that to some of the problem users and see if anything different happens. I'm also wondering if this was happening before, but your other problems w/ospf were "masking" the problem. What version of arc? C -- Charles Sprickman spork@inch.com On Mon, 13 Jan 2003, Joshua Coombs wrote:
After converting our TC network over to rip, things have gone quiet on the dialup front. ISDN multichanel 'just works' and wierd breaks in routing have been pleasently absent from my list of tasks to debug. Unfortunatly, I bragged about this too much and a new issue has popped to the surface.
It appears that for quite awhile (as far back as I have logging) we've been handing out 0.0.0.0 as an IP to approx 1% of all connections. We have 3 /24's setup as dynamic public pools on each of 8 TC's. Radius has no knowledge of the ip pools, and as such doesn't specify one to use for connections. At PEAK capacity I show the pools running at about 2/3rds capacity (200 out of 254 in use). Customers auth, and recieve the all 0's ip, hang around for about 30 seconds, and disconnect. No errors are logged by the TC's or our radius server.
I'm stumped. This affects just about every type of caller we have, generic analog dyanmic to ISDN with static entries. I'm open to suggestions and appreciate any you can offer.
-- Joshua Coombs GWI Networking
_______________________________________________ USR-TC mailing list USR-TC@mailman.xmission.com http://mailman.xmission.com/cgi-bin/mailman/listinfo/usr-tc