ced@bcstec.uucp (Charles Derykus) (05/16/91)
Our caching only nameserver running on an RS6000, rev 3005 has started to break intermittently. It is also a resolver. /etc/resolv.conf is listed below: nameserver 127.0.0.1 nameserver 128.207.254.223 nameserver 128.207.254.44 nameserver 136.240.1.21 domain ca.boeing.com When the nameserver breaks, it stops resolving from the local nameserver cache and looks only to /etc/hosts. It also refuses to query other nameservers. The really weird thing is that it seems to get well after an indeterminate amount of time. Suddenly, the local nameserver starts resolving, apparently none the worse for wear. Sometimes I can stop/restart named to make it well, but this doesn't always work. Has anyone seen this or have any theories? Any help greatly appreciated. Charles DeRykus Internet: ced@bcstec.boeing.com Boeing Computer Services UUCP: ...!uunet!bcstec!ced Renton, WA. M/S 6R-37 (206) 234-9223
jackv@turnkey.tcc.com (Jack F. Vogel) (05/16/91)
In article <856@bcstec.boeing.com> ced@bcstec.uucp (Charles Derykus) writes: | |Our caching only nameserver running on an RS6000, rev 3005 has started to |break intermittently. It is also a resolver. /etc/resolv.conf is listed |below: |nameserver 127.0.0.1 |nameserver 128.207.254.223 |nameserver 128.207.254.44 |nameserver 136.240.1.21 |domain ca.boeing.com | |When the nameserver breaks, it stops resolving from the local nameserver |cache and looks only to /etc/hosts. It also refuses to query other |nameservers. You're a bit confused here, at least terminology wise, the nameserver NEVER looks at /etc/hosts, neither does the resolver. It is only the routine gethostbyname() which does this, and normally only when a query to the nameserver fails to resolve an address. You also don't say WHY you believe it is "broken". How do you know it has quit resolving names, what are the visible symptoms, knowing that may help figuring out what is really wrong. >Has anyone seen this or have any theories? There just isn't enough info here for theories. What you might find useful is when you believe the nameserver is in this non-functional state, turn on its debugging log (read the man page on named for details), try some queries and then look at the log for what it is seeing. I suppose its possible that the named process has gone off to sleep() somewhere below PZERO and simply isn't responding, in which case you won't get a log file either, but that seems unlikely. Even better, would be to use nslookup (which IBM doesn't ship so you would have to 'roll your own') to directly interrogate the server when its in this state. For those unfamiliar with it, 'nslookup' is a utility provided in the Berkeley BIND distribution, anyone administering a network using a nameserver should have it around for debugging problems. Good Luck! Disclaimer: I don't speak for my employer or IBM. -- Jack F. Vogel jackv@locus.com AIX370 Technical Support - or - Locus Computing Corp. jackv@turnkey.TCC.COM
reilly@scotty.dccs.upenn.edu (G. Brendan Reilly) (06/02/91)
Actually, I just found that the 3005 nameserver will do a partial zone load and still use the data. When I teach the DNS course it is emphasized over and over not to do this. Am in the process of documenting this bug and submitting it to IBM. If you want to see the proof please send mail to reilly@sec.com.