David Luyer writes:
| The one current 'problem' with multicast in 1.1.1 which may not yet be
| resolved is that the cache is in its own multicast group, and hence
| responds to its own ICP queries.
How about this ... ?
Cheerio,
Martin
*** src/comm.c.orig Fri Feb 7 04:57:12 1997
--- src/comm.c Thu Feb 13 15:36:00 1997
***************
*** 1154,1159 ****
--- 1154,1160 ----
const ipcache_addrs *ia = NULL;
int i;
int x;
+ char c = 0;
for (s = Config.mcast_group_list; s; s = s->next) {
debug(5, 10, "comm_join_mcast_groups: joining group %s on FD %d\n",
s->key, fd);
***************
*** 1166,1171 ****
--- 1167,1177 ----
if (x < 0)
debug(5, 1, "comm_join_mcast_groups: FD %d, addr: %s [%s]\n",
fd, s->key, inet_ntoa(*(ia->in_addrs + i)));
+ x = setsockopt(fd, IPPROTO_IP, IP_MULTICAST_LOOP, &c, 1);
+ if (x < 0)
+ debug(5, 1,
+ "comm_join_mcast_groups: can't disable m'cast loopback: %s\n",
+ xstrerror());
}
}
#endif
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:34:26 MST