[Debian] Re: Bug#513310: vzctl fails to set capabilities, and subsequently fails to start any VE

[Daniel Pittman]

Kir Kolyshkin <kir at openvz.org> writes:

> This is caused by newer kernel headers (in this case on a build system
> that was used to build this vzctl package), and is fixed in
> vzctl-3.0.23. See the following git commit:

vzctl 3.0.23-2 is available in experimental, so I have installed it and
tested it on my machine; it addresses the problem and the VE will again
start.

> So the solution is either to upgrade to vzctl-3.0.23 or to backport
> this simple fix.

I can confirm that the newer package version resolves the problem.

> Ola Lundqvist wrote:
>> Hi Daniel
>>
>> This is interesting as it works very well on my systems. On other
>> hand that system is a 686 based one.
>>
>> You write that you have not significantly changed your system, but at
>> the same time you write that you are not sure that it has ever worked
>> with the 2.6.26 kernel.

Sorry, I see I was unclear: I have upgraded to sid, which significantly
changed the system, but the OpenVZ configuration remained stable.

I thought that the VE had started successfully under 2.6.26 before, but
could only confirm from my logs that I had used it under 2.6.24.

Sorry for being so unclear, and thankfully Kir has saved me by
identifying the problem despite my poor communication.

Regards,
        Daniel