[ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
Let's suppose we further modify our filter to also handle accounting requests. To discern between the authentication and accounting requests we'll prefix each authentication request with the word `auth' and each accounting request with the word `acct'. Furthermore, the input line for accounting requests will contain a timestamp.
Now, our filter program will look as follows:
#! /bin/sh AUTH_DB=/var/db/userlist ACCT_DB=/var/db/acct.db while read CODE NAME CLID DATE do case CODE auth) if grep "$1:$2" $DB; then echo "0 Service-Type = Login, \ Session-Timeout = 1200" else echo "1 Reply-Message = \ \"You are not authorized to log in\"" fi acct) echo "$CODE $NAME $CLID $DATE" >> $ACCT_DB done |
Its declaration in the `raddb/config' will also change:
filter check_clid { exec-path "/usr/libexec/myfilter"; error-log "myfilter.log"; auth { input-format "auth %C{User-Name} %C{Calling-Station-Id}"; wait-reply yes; }; acct { input-format "acct %C{User-Name} %C{Calling-Station-Id} %D"; wait-reply no; }; }; |
(The input-format
lines are split for readability. Each of them
is actually one line).
Notice wait-reply no
in the acct
statement. It
tells radiusd
that it shouldn't wait for the response on
accounting requests from the filter.