Subject: Re: [vserver] Problem with delete vserver with SYS_ADMIN set
From: Steve Kieu <msh.computing@gmail.com>
Date: Sat, 27 Aug 2011 11:42:24 +1000
Sat, 27 Aug 2011 11:42:24 +1000
All right, I edited the file etc/rc6.d/K07umountroot  and just commented out
all mount remount ro thing - this works

Thank you very much for your help. If someone know why it does not happen
with x86 I still want to hear :-)

cheers,



On Sat, Aug 27, 2011 at 11:34 AM, Steve Kieu <msh.computing@gmail.com>wrote:

> Confirmed this is the case. I tried to start in rescue like vserver  test
> start --rescue /bin/bash
>
> and then exit the bash shell, and I am able to delete it - this bypass all
> init scripts.
>
> But why it does not happen with x86 which is unknown to me now. The vserver
> root file system is created by debootstrap, but I have many other ported
> from openvz templates and manually build using rsync method - so I need to
> trace it down where the remount ro.
>
> Now I have to trace it doen where the actual remount as ro in the shutdown
> script :-( - any hint ?
>
> of course they are somewhere in /etc/rc6.0/ ... :-)
>
> many thanks
>
>
> On Sat, Aug 27, 2011 at 11:17 AM, Steve Kieu <msh.computing@gmail.com>wrote:
>
>> What I can not quite understand is that why it only happen with sparc
>> kernel. x86 and x86_64 same kernel same utils are fine.
>>
>> It might be that when the MOUNT permission are allowed inside guest, guest
>> startup script mount it as read - only. It is done in the guest  init system
>> (I mean sysv init script in /etc/init.d/ ) ? Or it  is controlled by vserver
>> utils?
>>
>>
>>
>>
>> On Fri, Aug 26, 2011 at 9:49 PM, Herbert Poetzl <herbert@13thfloor.at>wrote:
>>
>>> On Fri, Aug 26, 2011 at 12:48:21PM +0200, Grzegorz Nosek wrote:
>>> > W dniu 26.08.2011 12:24, Steve Kieu pisze:
>>> >> No speacial erroe at dmesg and message at all. The error
>>> >> message when running vserver delete is just thousand of lines
>>> >> like this
>>>
>>> >> remove `/var/lib/vservers/test/sbin/udevadm': Read-only file system
>>>
>>> > You probably have the guest shutdown scripts remounting all
>>> > filesystems into read-only mode.
>>>
>>> yeah, that sounds quite likely, please add some kind
>>> of mount info (probably best would be cat /proc/mounts)
>>> into the post-stop scriptlet ...
>>>
>>> something like:
>>>
>>> cat /proc/mounts >/tmp/mount.info
>>>
>>> if this is the case, your problem is a badly cleaned
>>> up guest system, which should be easy to fix
>>>
>>> best,
>>> Herbert
>>>
>>> > Best regards,
>>> >  Grzegorz Nosek
>>>
>>
>>
>>
>> --
>> Steve Kieu
>>
>
>
>
> --
> Steve Kieu
>



-- 
Steve Kieu


All right, I edited the file etc/rc6.d/K07umountroot  and just commented out all mount remount ro thing - this works

Thank you very much for your help. If someone know why it does not happen with x86 I still want to hear :-)

cheers,



On Sat, Aug 27, 2011 at 11:34 AM, Steve Kieu <msh.computing@gmail.com> wrote:
Confirmed this is the case. I tried to start in rescue like vserver  test start --rescue /bin/bash

and then exit the bash shell, and I am able to delete it - this bypass all init scripts.

But why it does not happen with x86 which is unknown to me now. The vserver root file system is created by debootstrap, but I have many other ported from openvz templates and manually build using rsync method - so I need to trace it down where the remount ro.

Now I have to trace it doen where the actual remount as ro in the shutdown script :-( - any hint ?

of course they are somewhere in /etc/rc6.0/ ... :-)

many thanks


On Sat, Aug 27, 2011 at 11:17 AM, Steve Kieu <msh.computing@gmail.com> wrote:
What I can not quite understand is that why it only happen with sparc kernel. x86 and x86 64 same kernel same utils are fine.

It might be that when the MOUNT permission are allowed inside guest, guest startup script mount it as read - only. It is done in the guest  init system (I mean sysv init script in /etc/init.d/ ) ? Or it  is controlled by vserver utils?




On Fri, Aug 26, 2011 at 9:49 PM, Herbert Poetzl <herbert@13thfloor.at> wrote:
On Fri, Aug 26, 2011 at 12:48:21PM +0200, Grzegorz Nosek wrote:
> W dniu 26.08.2011 12:24, Steve Kieu pisze:
>> No speacial erroe at dmesg and message at all. The error
>> message when running vserver delete is just thousand of lines
>> like this

>> remove `/var/lib/vservers/test/sbin/udevadm': Read-only file system

> You probably have the guest shutdown scripts remounting all
> filesystems into read-only mode.

yeah, that sounds quite likely, please add some kind
of mount info (probably best would be cat /proc/mounts)
into the post-stop scriptlet ...

something like:

cat /proc/mounts >/tmp/mount.info

if this is the case, your problem is a badly cleaned
up guest system, which should be easy to fix

best,
Herbert

> Best regards,
>  Grzegorz Nosek



--
Steve Kieu



--
Steve Kieu



--
Steve Kieu