From: John P. Eisenmenger (jpe_at_eisenmenger.org)
Date: Sun 03 Nov 2002 - 22:45:25 GMT
On Sun, 3 Nov 2002, Matt Ayres wrote:
> On 3 Nov 2002, Klavs Klavsen wrote:
> > On Sat, 2002-11-02 at 23:36, Matt Ayres wrote:
> > >
> > > Did you try this first before making assumptions?
> > sure did, otherwise I wouldn't have known what fakeinit did.
> Try again. This is what a vps -axuw | grep init returns on my host
Hmm. in v .21 of the vserver script the "fakeinit causes STARTCMD to be,
literally, "/sbin/init". This does not work for my Gentoo system, though
I think mine is based on Gentoo 1.2 and not 1.4rc1. My system merely
generates a usage message for init indicating that is expects a runlevel
designation on the command-line. Nothing gets started.
But, as I've said before, I think getting Gentoo working is a specific
symptom of a more general problem. The general problem is that the
vserver script makes certain assumptions about the structure of the
vserver that won't always be true.
I mean, who knows if the next great distribution will even use System V
start scripts? Maybe it'll use the BSD setup or even, God forbid, the old
HP/UX startup methods. Or maybe it'll have an entirely new method coded
in some language du jour (forgive my spelling please :-)...
Attached you will find my patch to the vserver script. It recognizes 2
new values in the vserver.conf file: S_START and S_STOP. Specifying these
values will override the vserver script's current logic (but it will not
bypass the "fakeinit" flag). In this way the current functionality of the
script is preserved and an option is provided to cleanly handle "weird"
FWIW my gentoo server's configuration file looks something like this now:
ULIMIT="-H -u 1000"
So when I start my Gentoo vserver, the vserver script executes "/sbin/rc
default" in its context. When I stop it, "/sbin/rc shutdown" is executed.
This seems to work well for me, and I hope others will find this
-- John P. Eisenmenger jpe_at_eisenmenger.org