bind-8.2.2-P5 and solaris 2.5.1

Kevin Darcy kcd at daimlerchrysler.com
Fri Jun 16 18:37:05 UTC 2000


sonam at escape.com wrote:


> [ls -l of libraries and devices nodes, snipped for brevity]
>
The device node major/minor numbers match mine exactly, but the nodes are only
writable by root/other; what ID is "named" running under and does it have write
permission to those nodes?

The libraries are quite similar to mine; the minor differences are probably
because we are at a different patch level.

I found "setnetconfig" in the "nsl" library, and a reference to it in the "socket"
library. It appears that some of the low-level socket calls invoke "setnetconfig"
for some reason. The fact that it is failing implies that there is something
screwy with your netconfig configuration in the chroot jail. Do you have an
/etc/netconfig in there?


- Kevin

> On Fri, 16 Jun 2000, Kevin Darcy wrote:
>
> > sonam at escape.com wrote:
> >
> > > Hi Kevin,
> > >
> > > When I run named "chroote'd" I get an error message saying
> > > socket(SOCK_RAW): Bad file number
> > > Abort - core dumped
> >
> > Check all of the device nodes in your chroot jail -- permissions, major/minor
> > numbers, etc. Named appears to be failing when trying to open a socket to
> > probe your interfaces. (Question to developers: since the socket call uses
> > SOCK_DGRAM, why do the error messages report SOCK_RAW?).
> >
> > > In the log message it says,
> > >
> > > Jun 16 13:33:13 server named[2604]: _s_match_netconf: setnetconfig failed
> > > Jun 16 13:33:13 server named[2604]: ctl_server: socket: Bad file number
> > > Jun 16 13:33:13 server named[2604]: _s_match_netconf: setnetconfig failed
> > > Jun 16 13:33:13 server named[2604]: socket(SOCK_RAW): Bad file number
> > > Jun 16 13:33:13 server named[2604]: socket(SOCK_RAW): Bad file number
> >
> > Very odd. I find no references to _s_match_netconf or setnetconfig in the
> > BIND source or in my Solaris 2.5.1 named binary. Are you certain you didn't
> > get some bogus library linked in there somehow?
> >
> >
> > - Kevin
> >
> >
> >
> >
> >






More information about the bind-users mailing list