About this list Date view Thread view Subject view Author view Attachment view

From: Liam Helmer (linuxlists_at_thevenue.org)
Date: Wed 12 May 2004 - 00:26:48 BST


I think that this is because bind9 uses linux-capabilities to do it's
change to a particular user -> thus, when capabilities aren't present in
the compilation, no user switch is possible. That's been my take on it
at least.

I'm in the process of switching my sites to pdns anyways though -> which
removes the problem a different way.

Cheers,
Liam

On Tue, 2004-05-11 at 21:59, Herbert Poetzl wrote:
> On Tue, May 11, 2004 at 11:50:16PM +0200, Mauro Calderara wrote:
> > Hi everyone
> >
> > i'm not sure whether i'm posting to the correct list but i expect the
> > most people here that have seen the problem i met:
> >
> > i'm using vserver on debian woody (2.4.25-vs1.26) and Paul Sladen's
> > bind9-nocapset stuff from his page. Has anyone ever tried to run named
> > of this package as non-root user using the -u option? i get here
> >
> > named: -u not supported on Linux kernels older than 2.3.99-pre3 or
> > 2.2.18 when using threads
> >
> > when starting named. my options are "-u named -t /var/bind9 -c
> > /etc/bind/named.conf"
> > ... it is to say that this problem occurs both inside the vserver and
> > on the host-os, i.e. it is in fact not a vserver-problem. I still ask
> > on this list because if there are people using this package, they will
> > probably run vservers. I tested another package with the same result.
> > This was an issue 2 years ago on this mailing list but the guy with the
> > problem solved it by switching to djbdns which i'd like to avoid. Has
> > anyone seen this problem too? I have it here on 2 different machines
> > but according to google it seems not a problem for lots of people,
> > there are only few hits and no fixes or explanations i found ...
> > Bertl, the good-karma-guy from #vservers on OFTC suggested i use strace
> > to investigate the problem but i was unable to find something useful.
> > maybe bind has a problem with the "vs-1.26"-part of the output ...
> > could that be? i.e. how could i hack the kernel-source to return e.g.
> > 2.4.25 instead of 2.4.25-vs1.26 ... or is that a bad idea because it'd
> > break other stuff? the strace-output is available on
> > https://limes2.dnsalias.net/namedstrace .
>
> The following might sound funny, but I'd suggest to
> give it a try:
>
> rename the kernels version/extraversion to something
> like 2.4.25 instead of 2.4.25-vs1.26 and try again,
> maybe the bind9 package is just to stupid to parse
> the kernel version correctly ...
>
> (although I doubt this, and I didn't have a look at
> the bind source code yet ...)
>
> HTH,
> Herbert
>
> PS: this can be done in the toplevel Makefile ;)
>
> > thanx in advance for any hints on how to solve the issue and have a
> > nice day everyone
> >
> > Mauro
> >
> > _______________________________________________
> > Vserver mailing list
> > Vserver_at_list.linux-vserver.org
> > http://list.linux-vserver.org/mailman/listinfo/vserver
> _______________________________________________
> Vserver mailing list
> Vserver_at_list.linux-vserver.org
> http://list.linux-vserver.org/mailman/listinfo/vserver
>

_______________________________________________
Vserver mailing list
Vserver_at_list.linux-vserver.org
http://list.linux-vserver.org/mailman/listinfo/vserver


About this list Date view Thread view Subject view Author view Attachment view
[Next/Previous Months] [Main vserver Project Homepage] [Howto Subscribe/Unsubscribe] [Paul Sladen's vserver stuff]
Generated on Wed 12 May 2004 - 21:12:55 BST by hypermail 2.1.3