smart@manta.mel.dit.csiro.au (Robert Smart) (07/25/90)
I hope these matters are unrelated! 1) I strongly favour use of MXes being optional in sendmail. I.e. you should be able to force some mail to go a particular route. For example the 5.64+IDA package has an option for a RELAY_HOST. But actually you could end up sending mail meant for the RELAY_HOST to a secondary MX of the RELAY_HOST, which might not be the same thing. My previously offered code for this was to only do MX if the $h was left off the [IPC] parameters in the M line. This had the mysterious effect that sendmail in daemon mode would not try to deliver to MX destinations when running the queue. Running the queue by hand worked fine. So instead I now have [IPC] and [IMX] mailers and the diff looks like this: *** deliver.c Sun Jul 15 20:44:32 1990 --- deliver.c~ Fri Jul 13 14:24:46 1990 *************** *** 436,447 **** { rcode = EX_OK; #ifdef NAMED_BIND - #ifdef OPTMX - if (host[0] && host[0] != '[' && - strcmp( m->m_mailer, "[IMX]") == 0) - #else if (host[0] && host[0] != '[') - #endif { expand("\001w", buf, &buf[sizeof(buf) - 1], e); Nmx = getmxrr(host, MxHosts, buf, &rcode); --- 436,442 ---- *************** *** 797,803 **** return (0); } ! if (strcmp(m->m_mailer, "[IPC]")==0 || strcmp(m->m_mailer, "[IMX]")==0) { #ifdef HOSTINFO register STAB *st; --- 792,798 ---- return (0); } ! if (strcmp(m->m_mailer, "[IPC]") == 0) { #ifdef HOSTINFO register STAB *st; ----------------------------------------------------------------------- And this seems to work fine. 2) I built my sendmail on a SunOS 4.0.3 machine with BIND 4.8.2 (beta?) libraries. Later I installed it on a SunOS 4.1 machine with default Sun resolver library running Yellow Pages (NIS?). This had the effect that incoming SMTP calls from a couple of hosts would make the receiver process go into an infinite loop chewing up cpu time. The hosts in question did not have reverse lookups for their IP adresses and indeed the names they report on the HELO line are not in the DNS. I installed the sendmail binary linked on my 4.0.3 machine and the problem went away. Any suggestions would be welcome. Bob Smart <smart@mel.dit.csiro.au>