On Thu, Oct 18, 2007 at 07:01:59PM -0700, Richard A Nelson wrote: > On Thu, 18 Oct 2007, Kees-Jan Dijkzeul wrote: > > This whole mess was concocted by me and a few others back when we were > forced to live on PPP/dialup or DHCP dsl/cable - and it worked just fine > for us, but at least I always ran a split horizon/view local DNS on my > mail servers - and thus never really ran into this issue.
I manged to run for years without it being a problem. I don't know what happened that I got bitten that time either. The system serves DHCP and has static ip addresses. My fault for having dialup.m4 referenced. > I found a bug in the dynamic script that can cause early termination > of the host lookup loop - and explains the ';;....' edit. > > I'm going to fix that, but before I put it out, I would like your > opinions on how to handle the remaining issue: > > /usr/share/sendmail/dynamic::update_host is called when the IP has > changed, and it calls find_host to do the resolution... just blindly > assuming it has done the job properly. > > If find_host works, there is no problem... > > If it fails, however, there are some options - none good ;( > > *) Keep sendmail from starting in an unknown state, and invent > a means to do the needed DNS lookup during > /etc/init.d/sendmail start > This likely means actually keeping some state around - and > we could even retry later on DHCP RENEWs ! > > *) Keep sendmail the 'logically offline state' - mail will > not be sent, I will have to investigate if it will accept > mail in this state or not (doing so could be *bad*) > > *) Use the prior hostname, which *may* be invalid - potentially > causing bounces, DNSBL additions, etc. I would say a modified last one. Keep the current configuration and sendmail running, but invalidate the peer and IP address cache, so that they don't have to change for the script to try again. Though I was looking through the /usr/share/sendmail/dynamic script I have and I don't see how it compares the current ip address to the last one. > -- > Rick Nelson > I stopped a long time ago to try to find anything in the bug list of dpkg. > We should run for an entry in the Guinness Book of Records. > -- Stephane Bortzmeyer -- David Fries <[EMAIL PROTECTED]> http://fries.net/~david/ (PGP encryption key available) -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
