Re: [Vserver] problems starting and stopping a vserver: partially solved

From: alexander goeres - lieblinx.net <agoeres_at_lieblinx.net>
Date: Fri 03 Mar 2006 - 13:22:24 GMT
Message-Id: <200603031422.24419.agoeres@lieblinx.net>

Hello Herbert,
I found my error with the zope-start at vserver-startup: it was a missing PATH
in the vserver's init script for zope. I'm sorry to having bothered the list
and you with that..

That doesn't exactly solve my other problems regarding the unclean shutdowns
of a vserver that was started with the fakeinit-flag. But it changes them
more or less to purley academic problems, since I don't need this fakeinit
stuff now no more.

best regards,
Alexander

Am Freitag, 3. März 2006 13:12 schrieb alexander goeres - lieblinx.net:
> Hello Herbert,
>
> Thanx for the quick reply.
>
> Am Donnerstag, 2. März 2006 16:08 schrieb Herbert Poetzl:
> > On Thu, Mar 02, 2006 at 03:39:09PM +0100, alexander goeres - lieblinx.net
>
> wrote:
> > > Hello List ,
> > >
> > > I recently set up a new machine with Version 2.0.1 of the vservers and
> > > 0.30.210 of the util-vserver. It's a vanilla kernel 2.6.15-3 on a
> > > debian host on a AMD 64 3200+.
> > >
> > > There I discovered some problems with the startup of services inside
> > > the vservers (debian guests). Zope instances that are started through
> > > init-scripts fail and die after some time without any further notice.
> >
> > might be interesting to add an 'strace -fF -o zope.trace ...' to
> > the relevant section of the startup script (the zope script)
> > so that we might get an idea what happens and why it fails
>
> I followed your advice and got about 18MB of logfile for the start
> procedure during vserver startup and about 6Mb during a start from within a
> vserver. Well, now I'm trying to pin down an inhouse programmer to explain
> that to me .. :-/
>
> > > Starting them after the init-procedure works as it should. I fear that
> > > this is mainly zope-related but it can't be only a mistake with zope
> > > since it was working perfectly on older vserver-versions.
> >
> > do they stay 'forever' then, or do they fail/die after some time
> > too, when started _after_ the init procedure?
>
> Starting zope with the init-script after the vserver-startup makes the
> instances permanent, just as it should.
>
> > > If anybody here had run into a similar problem and had solved it I'd
> > > gladly accept any hints to the solution.
> >
> > are you sure that the 'only' thing you changed was the kernel and
> > util-vserver? maybe a distro update too? maybe a new hardware?
>
> I was mistaken to claim that the zope-startup through init-scripts worked
> on previous vserver versions. I checked all our hosts and found that it had
> _never_ worked on the hosts with vserver 1.9.x to 2.0.1. I just never had
> the idea to test that, assuming that of course it would be functional and
> since I never had to restart the vservers I failed to notice this.
>
> So I guess it's rather really a zope problem. I put that in the back to
> investigate it later. So there's only my problems with using fakeinit left.
>
> > > Trying to solve this I started the relevant vservers with the
> > > fakeinit-flag and the plain initstyle. Now the start of the vserver's
> > > init scripts is
> > >
> > > normal but stopping the vserver fails with the notice:
> > > >host:/etc/vservers# vserver testserver stop
> > > >A timeout occured while waiting for the vserver to finish and it will
> > > >be killed by sending a SIGKILL signal. The following process list
> > > >might be useful for finding out the reason of this behavior:
> > > >
> > > >----------------------------------------------------------------------
> > > >12105 120 testserver ? SNs 0:00 init [2]
> > > >12254 120 testserver ? SNs 0:00 /sbin/syslogd
> > > >12261 120 testserver ? SNs 0:00 /usr/lib/exim/exim3 -bd
> > > > -q30m 12266 120 testserver ? SNs 0:00 /usr/sbin/inetd
> > > >12273 120 testserver ? SNs 0:00 /usr/sbin/sshd
> > > >12276 120 testserver ? SNs 0:00 /usr/sbin/atd
> > > >12279 120 testserver ? SNs 0:00 /usr/sbin/cron
> > > >----------------------------------------------------------------------
> > >
> > > Sometimes this list is empty.
> >
> > that's not too unusual, and you might fix that in the shutdown
> > script by killing off all kind of stuff inside the guest, which
> > just seems not to be done atm ...
>
> Starting and stopping the vservers with the sysv initstyle and without the
> fakeinit flag shows that the services are started and stopped as intended.
> All services are started though, when starting a vserver with the fakeinit
> flag. Why should the shutdown procedure fail?
>
> Is there anything to be changed within the vservers when switching to plain
> initstyle and using the fakeinit flag? The vservers themselves are all
> debian guests.
>
> > > I'd be glad if anybody had an idea how to debug this: either start the
> > > services nomally at vserver-startup (preferred) or getting rid of this
> > > fakeinit-stop problem.
> >
> > HTH,
> > Herbert
>
> Regards,
> Alexander

-- 
netzwerk- & systemadministrator
-------------------------------------------
agoeres at lieblinx. net
tel.: +49 (0)30 / 61 20 26 87
fax: +49 (0)30 / 69 00 46 03
-------------------------------------------
lieblinx.net
     we do software
-------------------------------------------
reichenberger straße 125
10999 berlin
http://lieblinx.net
-------------------------------------------
_______________________________________________
Vserver mailing list
Vserver@list.linux-vserver.org
http://list.linux-vserver.org/mailman/listinfo/vserver
Received on Fri Mar 3 13:17:49 2006
[Next/Previous Months] [Main vserver Project Homepage] [Howto Subscribe/Unsubscribe] [Paul Sladen's vserver stuff]
Generated on Fri 03 Mar 2006 - 13:17:52 GMT by hypermail 2.1.8