[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Request for testing: /run and initscripts

On Wed, Apr 13, 2011 at 03:35:24PM +0200, Adam Borowski wrote:
> On Wed, Apr 13, 2011 at 02:24:30PM +0100, Roger Leigh wrote:
> > On Wed, Apr 13, 2011 at 03:20:38PM +0200, Adam Borowski wrote:
> > > On Wed, Apr 13, 2011 at 01:49:15PM +0100, Roger Leigh wrote:
> > > > I have now implemented this (though it's not the default).
> > > > 
> > > > I would very much appreciate it if anyone could take the time to
> > > > install the new initscripts and test it out.
> > > > 
> > > > http://people.debian.org/~rleigh/run/sysvinit_2.88dsf-13.3.dsc
> > > > http://people.debian.org/~rleigh/run/initscripts_2.88dsf-13.3_amd64.deb
> > > 
> > > It breaks down at least on vservers (which can't do mount() calls):
> > > 
> > > find: `var/run': No such file or directory
> > > fakerunlevel: open("/var/run/utmp"): No such file or directory
> > 
> > When is this, in postinst or init scripts?  We have logic
> > in postinst to detect chroot environments and not do any
> > mounting; could we add a similar check for vservers?
> postinst reported success.
> This error was upon trying to [re]start the vserver afterwards, not sure
> where exactly it comes from.

Not knowing anything about vservers, do they normally run
the standard rcS init scripts?  While this patch does change
some mounts to new locations, it's pretty much doing the same
thing as before.  If /dev/shm and /lib/init/rw aren't mounted,
it looks like they are not run.

After this failure, does /var/run exist?  Is it a directory or
symlink?  What's mounted there or what does it point to?  Is
that location also valid?

When the postinst ran, did it set up bind mounts, or did it
run the chroot setup only?  If vservers never run the rcS scripts,
we need to ensure that the package runs the same postinst codepaths
as for chroots.  Is it possible to determine if we are running in
a vserver enviroment in the postinst?

> You might want to ask someone who deals with LXC to test it as well, as
> it has a different approach to mounting filesystems inside the guest.

Looks a bit more comprehensive, and you have the choice of
running the native init.  We would need to know if rcS will be
run or not when running the postinst.  Comments from anyone using
lxc with Debian would be most helpful.

  .''`.  Roger Leigh
 : :' :  Debian GNU/Linux             http://people.debian.org/~rleigh/
 `. `'   Printing on GNU/Linux?       http://gutenprint.sourceforge.net/
   `-    GPG Public Key: 0x25BFB848   Please GPG sign your mail.

Attachment: signature.asc
Description: Digital signature

Reply to: