Re: DNS problem after upgrade to woody
kmself@ix.netcom.com writes:
> on Thu, Nov 30, 2000 at 10:51:02AM -0500, Giulio Morgan (mickle@panix.com) wrote:
> > kmself@ix.netcom.com writes:
> >
> > > on Wed, Nov 29, 2000 at 10:54:37PM -0500, Giulio Morgan (mickle@panix.com) wrote:
> > > > After upgrading, I am unable to connect to mailserver with
> > > > fetchmail, to remore hosts with apt, telnet, ssh, irc, lynx or
> > > > mozilla. Messages are:
> > > > unable to resolve host...
> > > > something wicked happened...
> > > > domain name could not be confirmed...
> > > > it appears that there is a problem with DNS and I have no idea how
> > > > to resolve it.
> > >
> > > Is your named daemon running? I've found that several daemons fail to
> > > restart after upgrades. DNS resolution tends to be poor without a
> > > running named....
> >
> > Thank you Karsten,
> >
> > I can't determine if my "named" is running.
>
> BTW, I should mention DNS isn't my strong suite, though I've poked at it
> enough for it to roll over and die from time to time.
>
> Q'n'D:
>
> $ ps aux | grep [n]amed
> root 1398 0.0 0.7 2612 1832 ? S Nov29 0:12 /usr/sbin/named
>
> You may only be using /etc/resolv.conf, you don't need to run named, and
> quite probably shouldn't.
>
> If you've got a file /etc/init.d/named, you can try running it:
>
> $ /etc/init.d/named start
>
> ...to see what happens. It'll just bitch at you if it's already
> running.
>
> > I checked logs to see if I
> > could find any meaningful indication, and comparing pre- and
> > post-upgrade logs, I noted that in the post-upgrade syslog, there is
> > an entry:
> > ...inetd[1731]: smtp/tcp: bind: Address already in use
>
> I believe this is different. named doesn't run out of indetd. I think
> it's trying to bind to a socket.
>
> > which does not appear in my pre-upgrade logs
> >
> > In the ppp.log, there is a line:
> > ...ppd[324]: not replacing existing default route to tap0 [0.0.0.0]
> > and then further down, after immediately after connection is established, the
> > following lines appear:
> > ...ppd[324]: local address...
> > ...ppd[324]: remote address 198.6.71.8 (correct address, line not copied
> > verbatim)
> > ...ppd[324]: script /etc/ppp/ip-up started
> > up to here, everything as before, then,
> > ...diald[207]: calling site 192.168.0.2 (don't know whose address)
> > ...diald[207]: Error opening device /dev/modem: no such file or directory
> > ...diald[207]: No devices free to call out on
>
> Are you getting your network connection? If you're relying on your
> ISP's DNS servers, and can't raise your PPP link, you're going to have
> trouble resovling hosts, but you can't reach them anyway <g>.
>
> Looks like you might have a modem config issue. Hardware layer beats IP
> layer any day of the week.
>
> > In pre-upgrade log, the next line after script /etc/ppp/ip-up started,
> > I don't remember exactly, but some indication of the successful
> > completion of the script, diald not invoked.
>
> Haven't used diald myself, but you might look at / play with pppconf.
>
>
> I'd strongly suggest looking at the Network Administrator's Guide, by
> Olaf Kirche, included as 'NAG' in /usr/doc/ldp-nag It's got a nice
> little section which should be all you need to get DNS configured for a
> standalone box, plus data on other networking issues. The new edition
> (released this year) is well worth the upgrade. There's another
> O'Reilly book on DNS and BIND, but it's most likely overkill.
>
>
> For troubleshooting:
>
> - Find out if you've got a network connection. Try pinging a fixed
> IP. For example, www.sourceforge.net (216.104.232.234 -- so you
> don't have to resolve it). If you can reach that, you've got an
> external link.
>
> - Check your DNS configs. It's usually something stupid like having
> 'order hosts,bind' out of order in /etc/host.conf (you want it as
> stated), the wrong DNS forwarders in /etc/resolv.conf, or a broken
> DNS config. I'd shoot for a caching-only nameserver first, before
> getting too tricked out.
>
> - Read the book, 'coz I'm all out of trix.
Thank you Karsten for all the advice. After spending some time trying to
configure a DNS server, trying to reconfigure resolve.conf, hosts.conf (both of
which were fine), and trying to get diald (which I had never explicitly tried
to iinstall) working, I finally dpkg removed diald, and just using my old
pppconfig, everything worked fine.
I guess that what was happening was not a "DNS" or "named" problem, just that
diald was initiated at boot, and then, after I made a ppp connection (pon),
each time I tried to connect to a remote host, diald attempted to make a
connection, and, finding the port and modem already in use, caused the
application to return a "standard" error message. Not having intended to
install diald during the upgrade, I didn't think out the error properly.
The only useful question for anyone else might be why did diald get installed
and initiated in the first place?
Thanks again,
Peter
--
Giulio Morgan
mickle@panix.com
Reply to: