named-xfer processes go defunct

[Kevin Darcy]

Kingsley Tart wrote:

> Often, when doing a named.reload on one of our new secondary
> DNS we get defunct named subprocesses, yet other times it
> goes through fine.
>
> We're using Slackware 7.0 with Bind4 as a secondary DNS. I created
> /etc/named.boot as below (lines are not indented in live file):
>
>   directory /var/named
>   cache . cache
>
>   secondary    zone1.com    x.x.x.x    zone1.com.zon
>   secondary    zone2.com    x.x.x.x    zone2.com.zon
>   secondary    zone3.com    x.x.x.x    zone3.com.zon
>   secondary    zone4.com    x.x.x.x    zone4.com.zon
> etc
>
> (where zone1.com is the zone, x.x.x.x is the ip of the primary and
> zone1.com.zon is the name of the file in which to store zone info).
>
> When I do a named.reload, it starts replicating the zones and then
> I end up with defunct processes before it finishes:
>
> 27524 ?        S      0:13 /usr/sbin/named
> 27546 ?        Z      0:00 [named-xfer <defunct>]
> 27552 ?        Z      0:00 [named-xfer <defunct>]
>
> At this point it does not replicate any more domains until I kill
> the process and start again. However, when I do this and then
> restart with "ndc start" I get two mode defunct subprocesses
> almost immediately.
>
> Is this a common problem? The defunct processes have been
> known to sit around for 30 minutes before I killed the lot and
> restarted named.

Upgrade to BIND 8.2.2-p7. There have been literally hundreds of bugfixes
since BIND 4. It's likely that one or more of them fixes the problem
you're having.


- Kevin