bind9 hang

[Mark_Andrews at isc.org]

> Mark_Andrews at isc.org wrote:
> > 
> >> 
> >> bind9 seems to hang after a prolong/frequent network outage.
> >> you can still use it to resolve for the zone that it authorizes,
> >> but NOT for the names for elsewhere.
> >> today, my named (bind-9.2.1rc1, solaris 7, chrooted) hang after
> >> one hour Internet uplink down. When the link is back online, I
> >> can "nslookup" for local names, but a query for "www.google.com"
> >> or "www.yahoo.com" just hang. I did a truss of the "named" process
> >> to see what it is doing. not much info (see attached). maybe
> >> it got to an infinite loop?
> >> 
> >> btw, I've seen similiar hang problem with bind9 that comes with
> >> redhat linux 7.2. I run it as a cache only nameserver on the local
> >> host (laptop). After a long power save mode, named will usually
> >> quit responding. I have to restart named everything I am back
> >> from a standby power-save mode.
> >> 
> >> Any thoughts?
> >> Thanks!
> >> Frank
> > 
> >        I would be looking at network congestion.  The truss shows
> >        named making 558 queries (fd=7) and only receiving 29 answers.
> > 
> 
> As soon as I killed and restarted the named, everything started to work.

	Which then makes me think that the problem lies below named.
	This is a threaded build so you are exercising less well
	tested parts of the system libraries and kernel.

	Next time this happens can you please get a time stamped
	truss output and a corresponding time stamped packet trace.
	If you could let bind9-bugs at isc.org know where to fetch
	these from I'll have a look at them.

> I don't think it is a network congestion problem. How would "restart named"
> help solve the network congestion problem?
> Looks like named got into a weird state after a long network outage.
> Even though network is finally backup, somehow named can't recover
> itself and the only solution is to kill/restart it. 
> 
> Frank
> 
--
Mark Andrews, Internet Software Consortium
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: Mark.Andrews at isc.org