BIND 9.2.3 and zone transfers larger than 64MB

Mark Hennessy mhennessy at cloud9.net
Fri Aug 27 14:15:23 UTC 2004


When the process dies, I get the following:
Aug 27 09:30:47 <host> /kernel: pid 83125 (named), uid 0: exited on =
signal 11
(core dumped)

This problem has not happened before this particular zone file started =
to get
around 64MB and larger.  It does not look like a memory problem with the
server, it has over 1 GB of RAM to play with.

Why would I be given advice to move back to BIND 8 from others who have =
seen
the problem go away by going back to BIND 8?  Simply saying that the =
machine
does not have sufficient memory doesn't make any sense.

--
 Mark Hennessy

-----Original Message-----
From: Jim Reid [mailto:jim at rfc1035.com]=20
Sent: Friday, August 27, 2004 9:58 AM
To: Mark Hennessy
Cc: bind-users at isc.org
Subject: Re: BIND 9.2.3 and zone transfers larger than 64MB=20

>>>>> "Mark" =3D=3D Mark Hennessy <mhennessy at cloud9.net> writes:

    Mark> I get the following output in my logs:

    Mark> Aug 27 08:45:17 <host> named[83125]: transfer of =3D
    Mark> 'rbl-plus.mail-abuse.org/IN' from <mail-abuse.IP>#53: failed
    Mark> while receiving responses: out of memory

    Mark> My named process also dies every few hours or so.

What gets printed in the logs?

    Mark> Is there any recommended course of action to take to fix
    Mark> this that does =3D not involve going back to BIND 8?

Your name server is clearly running out of memory. Perhaps you have
artificially low resource limits on the amount of RAM/VM that named
can use? Aside from throwing more memory at the name server, you might
want to look at why it's running out. Perhaps there is some other
problem like a memory leak in the OS or your name server is serving
too many (large) zones for the resources it's been allocated? What are
the OS-imposed limits? How much RAM/VM does the box have? What other
processes are guzzling RAM and VM resources? Do you have any resource
limits imposed in the named.conf file?

The size of the zone transfer has nothing to do with the problem.
Loading a 64Mb zone file shouldn't be an issue either. BIND9 regularly
loads zones bigger than 1 Gb. However loading this 64Mb file could be
a problem if the OS won't let the name server use that much extra RAM
or VM. Which is what seems to be the case here.





More information about the bind-users mailing list