I run Bind 8.3.3 and exim 3.36, along with fetchmail 5.9.11 - I don't think I'm considerably different from most others here at something like that. Until yesterday morning when Zapping in overlay mode on display :0 gummed up Railroad Tycoon 2 on display :1, and then succeeded between them in trashing the framebuffer, everything was working wonderfully. That was the first reboot in five weeks, and when the system came back, it couldn't find "localhost". It could find hosehead.dyndns.org and its various morphs (hosehead.dyndns.ws and mlkahnt.dyndns.org,) and it had no problem with 127.0.0.1, 192.168.1.1 (eth1) and 192.168.1.3 (eth0), either. /etc/hosts was unchanged, but /etc/resolv.conf had been reduced to just two entries - my ISPs nameservers, each preceded by a blank line, leaving me to believe that some program is doing the rewrite on bringing up the network link (I suspect the culprit is "usepeerdns" in /etc/ppp/peers/dsl-provider, although that has been there for over seven months.) I editted /etc/resolv.conf to add: domain hosehead.dyndns.org nameserver 127.0.0.1 at the top of the file and reloaded bind - things were generally recovered, with one noticeable exception: fetchmail will not feed collected mail into exim, report to the logs this: Dec 18 04:06:00 hosehead xinetd[8398]: warning: /etc/hosts.allow, line 13: can't verify hostname: gethostbyname(localhost) failed Dec 18 04:06:00 hosehead xinetd[8398]: libwrap refused connection to smtp from 127.0.0.1 Dec 18 04:06:00 hosehead fetchmail[8390]: SMTP connect to localhost failed Dec 18 04:06:00 hosehead fetchmail[8390]: SMTP transaction error while fetching from kingston.net Dec 18 04:06:00 hosehead fetchmail[8390]: Query status=10 (SMTP) I can't telnet to SMTP either, as either "localhost" or 127.0.0.1, although hosehead.dyndns.org (and its variants) is fine. Line 13 of /etc/hosts.allow is: ALL: LOCAL so I am wondering (and haven't been able to track down appropriate documentation thus far) if the definition of "LOCAL" has been tweaked, either in some securing layer, or something elsewhere that I might have inadvertently, umm, buggered. I haven't been touching firewall rules, so I shouldn't (hopefully) be blocking access to exim from 127.0.0.1, but to be honest, at this point, I am rather lost where next to look. Part of the frustration is that "localhost" is working on other ports (http and ldap work, https, ftp and ssh don't - telnet doesn't because I don't run a daemon to serve it, so it better not work!) I would also note that my system didn't automagically generate an ssl certificate for itself when it restarted - the previous one had expired by that point. I'm suspecting that the confusion at startup over the system identifying itself might be part of the reason there, and the expired certificate at the time of a restart being another. I'm hoping that fixing this "localhost" problem clears up everything else, but after sorting out the rest, my brain hurts. Maybe I need a brain doctor (Doctor! Doctor! Are you the DOCTOR! My BRAIN hurts! No! The Brain in my HEAD!) Maybe I just need to borrow a few other brains to think of stuff that I should know to look at - hello other brains. -- Mark L. Kahnt, FLMI/M, ALHC, HIA, AIAA, ACS, MHP ML Kahnt New Markets Consulting Tel: (613) 531-8684 / (613) 539-0935 Email: kahnt@hosehead.dyndns.org
Attachment:
signature.asc
Description: This is a digitally signed message part