From: Herbert Poetzl (herbert_at_13thfloor.at)
Date: Sun 24 Nov 2002 - 13:34:39 GMT
On Sun, Nov 24, 2002 at 08:58:32AM +0000, Paul Sladen wrote:
> On Sat, 23 Nov 2002, Herbert Poetzl wrote:
> > On Sat, Nov 23, 2002, Ivo De Decker wrote:
> > > (http://www.13thfloor.at/VServer/). Some remarks...
> >
> > maybe I'll drop the CAP_QUOTACTL anyway, and allow
> > virtual root to administer the users per se.
>
> I would be tempted to agree. It would probably be easier to just give uid=0
> in the context, control over its /16 worth of quotas.
the idea was to _allow_ some virtual servers to administer
their quota settings, where other servers are hard set
from _the outside_ (physical server) ...
> > hmm, why would you like to turn it off?
>
> > > When the patch is applied, the uid/gid change of files inside a vserver
> > > happens automatically. It would be nice if this could be turned on/off.
>
> Is there a better way of chcontext() passing *flags* through than using
> capabilities. Another two example of where this is of use is:
>
> mangle/do-not-mangle localhost
> proxy/do-not-proxy reboot
>
> It would be very useful to have this as an /option/ although I don't really
> think of it as a capability (maybe I'm wrong). `/proc/sys' entries perhaps?
hmm, you mean to switch it on and off based on the
context (virtual server)? not generally on/off, or
am I wrong again?
> > yes of course, you can chown the files back to the root server (ctx 0)
> > so everything is fine again. but you are right, context quota only
> > makes sense with fixed context ids.
>
> We should get the automatic allocated ctx-ids to start at 1024, or even just
> 128, which will stop anyone getting `stung'/confused and ranting about it.
Jacques said he will do something similar in the
next release, I only hope the limit is well defined,
because otherwise it will be hard to check for ...
> > > A cosmetic bug in the quotatools patch:
>
> Another one: Does `statfs()' need patching to return the ctx-hard-quota as
> the vroot disk size?
I guess this is more complicated, because statfs
queries the _real_ filesystem, and does not know
anything about the vroot device :(
maybe another aproach will be required to solve this ...
best,
Herbert
> -Paul
> --
> Nottingham, GB