Thanks for the reply, Alan DeKok wrote: > > WA Support <[EMAIL PROTECTED]> wrote: > > If the detail file gets the STOP record, why don't radutmp and radwtmp > > get pruned? It is like parts of the code are not really communication > > with each other. > > They do. It's just that the detail module does *nothing* with the > request, except log it to a file. So it can log *anything*. > > The other files have specific requirements, and therefore don't know > what to do with some packets, when those packets don't contain the > information then need.
Like what information? I need specifics please about what information might be missing that would make rlm_radutmp not prune the radutmp file. The only condition that I can find in the code is if the terminal server can not be identified, in which case radutmp is trusted to see if an MPP session is active. Is this the only check before trusting radutmp? My radutmp and radwtmp files are not being pruned, and I need to know exactly what could cause this condition. My radius server is talking to a bank of Nortel CVXes. Obviously the STOP signal is coming back from the CVXes when the modem line changes state, since the detail file is being updated. It appears that all the necessary information comes back with the STOP packets. What is it that rlm_radutmp needs to prune radutmp that it is not getting, and how can I monitor this to see where the plot is falling apart? I am not too concerned about radwtmp, but radutmp is how simultaneous users is checked, right? And I am trying to control simultaneous logons. Currently, I have had to hack rlm_radutmp to ignore radutmp. Not the correct solution, but it is the only thing that I can do until I figure out waht is not happening correctly. Thanks, Murrah Boswell - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
