On May 15, 2006 12:45:05 PM -0700 David Bustos <David.Bustos at sun.com> wrote: > Quoth Frank Cusack on Mon, May 15, 2006 at 11:34:24AM -0700: >> I have a service that doesn't stay enabled. >> >> I compiled my own BIND named. I used the Sun manifest, and just changed >> the path to run my local copy. (I deleted and imported the manifest after >> the change.) When my named dies, either via pkill or svcadm disable, >> smf thinks it got disabled and doesn't restart it. When the Sun named >> dies, the same way, smf does restart it. Both processes seem to exit >> the same way. > > This looks like 6226796 (dns/server ends up being disabled on reboot).
Sure does. I can't see that bug report for some reason, but from the the files in patch 119783-01 it certainly seems like a bind thing. I'll look at the source changes. > What sources are you compiling from? isc.org 9.3.1. Any ideas (before I go looking through sources) why named has to exit a certain way? It certainly seems to exit(0) in response to SIGTERM (from my limited truss output). Also, how could I tell smf/svcs to give me more details? Having a problem here is one thing, not telling me what's going reminds me of the Mac. -frank
