From: Jacques Gelinas (jack_at_solucorp.qc.ca)
Date: Tue 06 Aug 2002 - 22:59:52 BST
On Sun, 4 Aug 2002 20:06:01 -0500, Herbert Poetzl wrote
>
> hi everyone!
>
> I encountered this problem by mere accident, as
> I upgraded some apache modules on my vservers,
> and the upgrade didn't reload the apache config,
> but instead restarted the apache server on the
> real machine ...
>
> vrpm (the script) executes
>
> rpm --root /vservers/<server> <arguments> ...
>
> which is perfect for packages without pre/post
> install/uninstall scripts interacting with
> running processes, and/or requiring more than
> a chrooted environment, but results in strange
> effects otherwise ...
>
> for example: executing 'killall -USR1 httpd'
> will for sure do a graceful restart of apache,
> but not for the apache within the vserver *G*
>
> one solution to this problem could be:
>
> - copy the rpm to /vservers/tmp/<rpm-package>
> - call vserver <server> exec rpm <arguments>
>
> (this is what I usually do)
>
> I think a better (read: more practical) solution
> could be to enter the virtual server context,
> reduce the capabilities, but stay within the current
> filesystem (no chroot), and then call the main
> rpm with the --root option ...
This is what the next release will do. I had experienced the same problem
and came to this conclusions. It is not reducing the capabiity though. Maybe
it should
---------------------------------------------------------
Jacques Gelinas <jack_at_solucorp.qc.ca>
vserver: run general purpose virtual servers on one box, full speed!
http://www.solucorp.qc.ca/miscprj/s_context.hc