bind 8.2.5 mysteriously hanging

[Mark_Andrews at isc.org]

> 
> We have two machines running Solaris 2.6, and bind 8.2.5.  One master
> for internal networks and responds to queries for outside
> destinations, the other a slave for internal networks - also queries
> for the outside world.  The slave much more often than the master, but
> named processes on both of these machine have been periodically
> hanging for about 30 minutes.  Here's a debug excerpt from one such
> hang:
> 
	When it happens again, attach a debugger to the process and
	get a stack backtrace.  Send the results to bind-bugs at isc.org.

	Mark
> 
>   -----------
>   resp: nlookup(225.217.130.157.in-addr.arpa) qtype=12
>   resp: found '225.217.130.157.in-addr.arpa' as '130.157.in-addr.arpa' (cname
> =0)
>   evSetTimer(ctx 0x1155f8, func 0x3795c, uap 0, due 1012492031.000000000, int
> er 0.000000000)
>   evSetTimer(ctx 0x1155f8, func 0x3795c, uap 0, due 1012492031.000000000, int
> er 0.000000000)
>   resp: forw -> [192.5.6.32].53 ds=4 nsid=33074 id=54171 34ms
> 
>   ========= HERE ==================
> 
>   Cleaned cache of 10225 RRsets
>   resend(addr=2 n=1) -> [152.228.34.1].53 ds=4 nsid=61027 id=54079 169ms
>   evSetTimer(ctx 0x1155f8, func 0x3795c, uap 0, due 1012492033.000000000, int
> er 0.000000000)
>   evSetTimer(ctx 0x1155f8, func 0x3795c, uap 0, due 1012492033.000000000, int
> er 0.000000000)
>   retry(0x372eca0): expired @ 1012492119 (2562 secs before now (1012494681))
>   -----------
> 
> Nothing abnormal is reported to syslog or /var/adm/messages - the
> process seems to take up the same amount of CPU and memory as when
> it's responding.  And after a while, it starts acting normally again.  
> A truss of the process shows *no* activity - no syscalls - and about a
> minute lag time beween sending the process kill signals and the
> processes receiving them, making them visible to truss.  
> 
> Any ideas where I should go from here?  We'll be upgrading soon, but
> I'd like to find out what's going on before we do.  
> 
> -fil
> --
> fil krohnengold
> network systems administrator
> american museum of natural history
> fil at amnh.org
> 
--
Mark Andrews, Internet Software Consortium
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: Mark.Andrews at isc.org