Intermittant resolving issue

Jim Reid jim at rfc1035.com
Sat Jul 1 02:25:49 UTC 2000


    >> I am experiencing an intermittant issue resolving a hostname on
    >> two of my nameservers. The other six resolve correctly, all of
    >> the time. This is what happens:

    >> .... complicated and not very useful discussion snipped ....

    >> Since I can get good answers consistantly on 6 servers, I
    >> assume that the gtei.net servers are serving up correct, timely
    >> and accurate answers. What I need to know is why the two
    >> servers get intermittant 'Servfail' on this hostname and how to
    >> fix it.

A SERVFAIL error is returned for a "server error". This typically
happens when the name servers for a domain are in a very bad state.
Another possibility for this error is when a loop is detected
i.e. Server A sends a query to B who thinks A should know the answer
already. See ns_forw.c. Resource record screwups can also cause
SERVFAIL: a too long CNAME chain or a domain that's missing glue.

    >> What should I be looking for in /var/adm?

Without knowing how all of those servers are configured, it's hard to
say. It's unlikely there will be anything meaningful in the name
server logs unless the debugging is on. Since you didn't supply the
names or addresses of these servers or the name that sometimes can't
be looked up, how can you expect someone on this list to troubleshoot
the problem? Maybe you're using forwarding. Maybe the 6 name servers
that have the answer have cached a long-lived TTL. Maybe they've got
the name as glue for some other zone that they serve? Who knows? Only
you and you're not telling....

Use dig to interrogate the name server(s) for the zone that contains
the failing name. Use it to check the delegation of that zone to those
servers. That should eventually identify where the problem lies. Pay
attention to whether the replies have the aa - authoritative answer
bit - set.



More information about the bind-users mailing list