Subject: Re: [vserver] Problems with scripts
From: Herbert Poetzl <herbert@13thfloor.at>
Date: Sun, 9 Sep 2007 00:59:23 +0200

On Fri, Aug 17, 2007 at 08:38:47AM -0300, Gustavo wrote:
> On 8/16/07, Herbert Poetzl <herbert@13thfloor.at> wrote:
> >
> > > > actually you do not need to care about unmounting it,
> > > > if you mounted it properly (inside the guest namespace)
> > > > as it will get unmounted when the context disappears
> > >
> > > Not really, it is unmounted but remains on the mount tab on vserver
> >
> > the mtab should/can be rewritten on guest startup
> > so that should not be a problem
> 
> I think that too, but for some reason in the host mount table the
> mount point of the loop is not updated though the loop is correctly
> unmounted on vserver stop, maybe I should mount it in any other place
> instead of initialize script, IŽll try on start/stop script with
> checking of a file on config dir.
> 
> > stop, and I also need to free the loop device after unmount.
> > >
> > > > Do you know any alternative for doing this?
> > > >
> > > > yeah, I would simply put it into the config fstab and
> > > > be done with it, the tolls will automagically mount it
> > > > for you and the kernel will take care of unmounting
> > >
> > > would it manage to free the loop?
> >
> > good point, not sure it will, personally I doubt it
> > but that could be a simple command in the shutdown
> > script, just to free up the device
> 
> 
> 
> yes, same as above, IŽll try on the vserver stop init script since
> postpost-stop doesnŽt works for this.
> 
> note: you do not have that issue with LVMs :)
> 
> 
> But I still need additional partitions :(
> I was trying to simulate the native scheme with an additional mount,
> so if you want to backup a vserver with it filesystem it shouldnŽt be
> much problem, actually I made some scripts for that too :)
> 
> Promise to share them with you all when they work fine.

any conclusions so far?
do we know why the fstab (in the guest config) 
doesn't work for this purpose?

TIA,
Herbert

> Thanks
> Gustavo