Re: [vserver] Help migrating from legacy: chbind: Segmentation fault

From: Herbert Poetzl <herbert_at_13thfloor.at>
Date: Wed 10 Mar 2010 - 11:39:08 GMT
Message-ID: <20100310113908.GK29490@MAIL.13thfloor.at>

On Tue, Mar 09, 2010 at 09:04:57PM +0100, Glomet wrote:
> Herbert,
> thank you very much:

> I've followed your instructions and it now works both on FC host and
> on Debian host.

good to hear!

> Just a problem remains:

> On both of the systems, new guest vservers now run fine.
> On the FC host (it is a Xeon based machine), old FC1 guest vservers
> images run fine, too.

> But on the Debian host (it is a AMD Athlon based machine), I wasn't
> able to run the same old FC1 guest vservers images that are now
> running on the FC host Xeon based machine.

> The error message follows, while starting the guest vserver:

> An error occured while executing the vserver startup sequence; when
> there are no other messages, it is very likely that the init-script
> (/etc/rc.d/rc 3) failed. [...]

try to start it with --debug and provide the output

> On the FC host Xeon based machine I fixed it just adding an "exit 0"
> at the end of the /etc/rc.d/rc in the guest vserver.

> But on the Debian host AMD based machine this doesn't work.

> Do you think this problem caused by the different CPU type?

could be, but rather unlikely ... i.e. it would mean that
at some point you compiled cpu specific code, which now
fails ...

> Maybe beacuse of compilation issues of the guest binaries?

guest binaries?

> Or am I missing something else?

I'd double check that you use recent util-vserver on
the system in question and that you do not start a 32bit
guest without the proper 32 bit personality

> Thank you very much.

best,
Herbert

> -- Umberto
>
> > On Mon, Mar 08, 2010 at 06:07:16PM +0100, Glomet wrote:
> >> Hello.
> >
> >> After years of succesfully usage of old versions of linux-vserver upon
> >> a few dedicate appliances, I now need to upgrade kernel because of new
> >> hardware requirements (and, I think, migrating from legacy version to
> >> the new one).
> >
> >> I have some problems: could any of you help me in fixing them?
> >
> >> I see at 13thfloor.at that the newest stable version is 2.2.0 against
> >> 2.6.20.4 kernel, but, unfortunately, at the time of that release
> >> of the kernel, there was a module for my ethernet card that wasn't
> >> reliable, and, since I cannot change the ethernet card with another
> >> one, I choose to use a newest kernel (but if you think I'm wrong with
> >> such this solution, please advice me).
> >
> >> After some searches, I choose to use 2.6.27.21 kernel.
> >
> > I'd suggest to go for the latest 2.6.31.x kernel, unless
> > 2.6.22.19 works for you (which has the latest 'stable'
> > kernel patch, as you can see on http://linux-vserver.org)
> >
> >> I had success, about an year ago, with 2.6.27.21-vs2.3.0.36.4 on a
> >> OpenSuse 11.1 host (util-vserver: 0.30.215): I know it was just an
> >> experimental version, but it run fine for about an year.
> >
> >> Unfortunately, using the same kernel / vserver / util-vserver versions
> >> on different systems, I have problems:
> >> Both on a Fedora Core 12 host, and on a Debian lenny host, the
> >> following error message appear:
> >
> >> deb-lenny:~# vserver webctl start
> >> /sbin/chbind: line 135: 4460 Segmentation fault "${create_cmd[@]}"
> >> "${chain_cmd[@]}" -- "$@"
> >
> > that looks very much like a miscompilation, I'd suggest
> > to get one of the recent pre-releases for util-vserver
> > and compile them with a known good toolchain
> > (do not use util-vserver 0.30.215 for recent kernels)
> >
> >> and then:
> >
> >> deb-lenny:~# vserver webctl status
> >> Vserver 'webctl' is stopped
> >> deb-lenny:~# vserver-stat
> >> CTX PROC VSZ RSS userTIME sysTIME UPTIME NAME
> >
> >> vserver-info output follows:
> >
> >> deb-lenny:~# vserver-info
> >> Versions:
> >> Kernel: 2.6.27.21-vs2.3.0.36.4
> >> VS-API: 0x00020304
> >> util-vserver: 0.30.215; Mar 8 2010, 17:21:34
> >>
> >> Features:
> >> CC: gcc, gcc (Debian 4.3.2-1.1) 4.3.2
> >> CXX: g++, g++ (Debian 4.3.2-1.1) 4.3.2
> >> CPPFLAGS: ''
> >> CFLAGS: '-g -O2 -std=c99 -Wall -pedantic -W
> >> -funit-at-a-time'
> >> CXXFLAGS: '-g -O2 -ansi -Wall -pedantic -W
> >> -fmessage-length=0 -funit-at-a-time'
> >> build/host:
> >> x86_64-unknown-linux-gnu/x86_64-unknown-linux-gnu
> >> Use dietlibc: yes
> >> Build C++ programs: yes
> >> Build C99 programs: yes
> >> Available APIs: v13,net,v21,v22,v23,netv2
> >> ext2fs Source: kernel
> >> syscall(2) invocation: alternative
> >> vserver(2) syscall#: 236/glibc
> >> crypto api: beecrypt
> >
> >> Paths:
> >> prefix:
> >> sysconf-Directory: ${prefix}/etc
> >> cfg-Directory: ${prefix}/etc/vservers
> >> initrd-Directory: $(sysconfdir)/init.d
> >> pkgstate-Directory: ${prefix}/var/run/vservers
> >> vserver-Rootdir: /vservers
> >
> >> Assumed 'SYSINFO' as no other option given; try '--help' for more
> >> information.
> >
> > testing with testme.sh is a good start, btw
> >
> >> I've tried some other kernel versions, each of them with the right
> >> vserver patches and util-vserver branch, but, on that kind of hosts
> >> (FC12 and Debian Lenny), I always had the same error message as above.
> >
> >> The guest vservers come from legacy, and they aren't updated (except
> >> for the exit 0 added to the /etc/rc in order to avoid the "An error
> >> occured while executing the vserver startup sequence; when there
> >> are no other messages, it is very likely that the init-script
> >> (/etc/rc.d/rc 3) failed." error message).
> >
> > should be fine
> >
> >> However, this solution was fine against the OpenSuse 11.1 host.
> >
> >> Where could I be wrong?
> >
> > most likely in the choice of tools ....
> >
> > best,
> > Herbert
> >
> >> Thank you
> >
> >> -- Umberto
> >
Received on Wed Mar 10 11:50:42 2010

[Next/Previous Months] [Main vserver Project Homepage] [Howto Subscribe/Unsubscribe] [Paul Sladen's vserver stuff]
Generated on Wed 10 Mar 2010 - 11:50:50 GMT by hypermail 2.1.8