Hi! > Really didn't notice that, happened couple of times, wonder why pppd > sends 2 acct-starts and sometimes with different session IDs. > Sorry to report this as it ain't a bug then, and thanks for the fast > reply.
Yes, problem in pppd's radius plug-in. Acct-Session-Id generated as strncpy(rstate.session_id, rc_mksid(), sizeof(rstate.session_id)); in radius_acct_start() function. rc_mksid defined in <pppd-source-root>/pppd/plugins/radius/radiusclient/lib/util.c as rc_mksid (void) { static char buf[14]; sprintf (buf, "%08lX%04X", (unsigned long int) time (NULL), (unsigned int) get pid ()); } i.e. when radius server don't answer to Acct-Start request pppd repeat it and generate another request Acct-Start, i.e. execute radius_acct_start() one more. so, i think, it's right to move line strncpy(rstate.session_id, rc_mksid(), sizeof(rstate.session_id)); from radius_acct_start() function to radius_init(), so rc_mksid() will called once. Mike. > On Tue, 2003-06-17 at 14:54, Chris Parker wrote: >> At 02:24 PM 6/17/2003 +0100, Manuel Sousa wrote: >> >Hi, all >> > >> >I've been using freeradius and noticed that sometimes the >> >Acct-Unique-Session-ID gave me different values for the same inputs. >> >A partial output of radiusd -X is: >> > >> >rlm_acct_unique: Hashing 'Acct-Session-Id = "3EEF21621014",User-Name = >> >"noc"' >> >rlm_acct_unique: Acct-Unique-Session-ID = "889e46aba4217ad4". >> > >> >rlm_acct_unique: Hashing 'Acct-Session-Id = "3EEF21631014",User-Name = >> >"noc"' >> >rlm_acct_unique: Acct-Unique-Session-ID = "6836c775ae8a6c48". >> > >> >Wonder if anyone else experienced the same problem. I'm using >> >freeradius-0.8.1. >> >> Look closer at the Acct-Session-Id, particulary the 8th position. Your >> first line has a '2', your second line has a '3'. They are not the same, >> hence the hash result is not the same. - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html