I'm a new subscriber, been busy reading the archives.

First off... I don't think I saw this patch go through which fixes a
problem exposed by the all-in-one patch:

*** hobbitd/hobbitd_alert.c.old Fri Aug 4 17:55:04 2006 --- hobbitd/hobbitd_alert.c Fri Aug 4 17:55:16 2006

*** 395,401 **** namelist_t *hinfo; int testdur = 0; FILE *logfd = NULL; ! activealerts_t *awalk = NULL;; int paramno = 0;

                     argi++; if (argi < argc) testhost = argv[argi];

--- 395,401 ---- namelist_t *hinfo; int testdur = 0; FILE *logfd = NULL; ! activealerts_t *awalk = NULL; int paramno = 0;

                     argi++; if (argi < argc) testhost = argv[argi];

I also found an off by one error in bbtest-net.c:

*** bbnet/bbtest-net.c.old Mon Aug 7 12:06:27 2006 --- bbnet/bbtest-net.c Mon Aug 7 12:07:28 2006

*** 1144,1150 **** } else { /* parent */ ! char ip[IP_ADDR_STRLEN];

             close(pfd[0]);
             pingcount = 0;

--- 1144,1150 ---- } else { /* parent */ ! char ip[IP_ADDR_STRLEN+1];

             close(pfd[0]);
             pingcount = 0;

IP_ADDR_STRLEN is defined to be 16, but in start_ping_service, it's
being used:

sprintf(ip, "%s\n", ip_to_test(t->host));

123.456.789.012 + \n + \0 is 17 characters. Whenever I had a
"full" IP address configured, bbtest-net would crash and burn.

The very simple patch above seems to have done the trick :)