On Fri, Mar 19, 2010 at 07:27:21PM +0100, Glomet wrote:
> Hello.
> Thanks to Herbert Poetzl's help, the chbind segmentation fault problem
> was fixed.
> But I'm still in troubles running old FC1 guest images inside the new
> linux-vserver.
> Kernel: 2.6.31.12-vs2.3.0.36.28
> VS-API: 0x00020305
> util-vserver: 0.30.216-pre2882
> On three different hosts:
> - same kernel release version on A, B, and C server;
> - same util-vserver release version on A, B, and C server;
> - different hosts linux distribution (FC6 on A, Debian on B, FC12 on C);
> - different CPUs (Xeon on A, AMD-64 on B, Xeon on C).
> A) FC1 guest runs: On a Xeon FC6 host: "vserver oldFC1guest start" runs fine
> B) FC1 guest doesn't run: On a AMD-64 Debian host: "vserver oldFC1guest
> start": "An error occured while executing the vserver startup sequence"
> C) FC1 guest doesn't run: On another Xeon host, FC12: "vserver oldFC1guest
> start": "An error occured while executing the vserver startup sequence"
did the guests run on the same machine before or did
you move them to a new one (maybe different arch)?
> Please, consider:
> - already added "exit 0" to the /etc/rc file in the guest
> - already manually typed "40000" in the /etc/vservers/oldFC1guest/context
> file in the host machine
> - ./testme.sh script doesn't report any problems
good start, what does 'vserver --debug <name> start'
and maybe with --debug-sysv too if appropriate?
> Please, consider more:
> - new tried debian guest images run fine on the AMD-64 debian based host
> machine (but not on the FC based xeons);
> - a new FC12 guest image seems to run on the Xeon FC12 based host machine,
> so I think vservers are running fine, the problem is just in migrating
> that guests from old kernels/legacy-vsevers to the new ones.
> Any suggestions?
try to configure the guest personality accordingly,
make sure to get single_ip special casing right and
avoid 'legacy' flags or similar in the config
HTH,
Herbert
> Thank you
>
> Regards
>
> -- Umberto
Received on Fri Mar 19 18:35:07 2010