[one-users] Stop-Resume/Migration Issue with checkpoint file in a bridged network NFS environment

[Javier Fontan]

It seems that you have apparmor enabled and it does not let libvirt
write checkpoint files. From OpenNebula documentation:

--8<------
If apparmor is active (by default it is), you should add $ONE_LOCATION/var
to the end of /etc/apparmor.d/libvirt-qemu

owner /path-to-one-location/var/** rw,
------>8--

Your "path-to-one-location/var" will be "/var/lib/one".

On Tue, Dec 28, 2010 at 5:53 PM, Jerry Deng <jerry at ideacore.com> wrote:
> Hope everyone is enjoy his/her holiday season.
> I am very much of a newbie here.
> With the following setup, I was able to deploy and run the VM (can ssh in
> it);  however, when trying shutdown VM, it tried and rebooted up.  (I
> already check ACPI="yes" on the template file).  When we do either
> stop-resume or migrate to another hosts (in which I have two hosts at the
> moment), it couldn't write the file to /var/lib/one/<vmid>/images (this is
> my system-wide express installed NFSed location).  So I went ahead and added
> to the front end server /etc/exports with "no_root_squash".  After which,
> the daemon is able to write checkpoint file to that location (filed will be
> owned by root with rw access given to root group and others).  However, the
> saving process never finishes, the KVM process is stucked with no logging
> information stored in any log, the VM will be in "SAVE" state and you could
> not apply any action to it using onevm command.   Then I came across a
> couple of blog posts where someone suggest adding a "touch" command to
> create the checkpoint file at the location as "oneadmin" (instead of "root")
> before the saving process begin.  With that dirty hack in the
> "one_vmm_ssh.rb" file, the checkpoint file will be an empty file owned and
> grouped by "oneadmin" with 0 byte in size.  Which probably means the saving
> process can't access the file at all.  Then I added "chmod 666" to the hack
> and that didn't work neither.  I am running out of ideas.  Please help,
> anyone, any ideas? inspire me!
> Setup:
> Ubuntu 10.04, KVM, NFS, OpenNebula Express 2.0 installed with bridged
> ethernet over eth0
> (virbr0 over eth0 where virbr0 is assign 192.168.0.201, and the hosted VMs
> are passed with public IP 192.168.0.220)
> Testing Image with TTYLinux.one template (mostly the same as the existing
> one listed here)
>
> Jerry Deng
>
> _______________________________________________
> Users mailing list
> Users at lists.opennebula.org
> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>
>



-- 
Javier Fontan, Grid & Virtualization Technology Engineer/Researcher
DSA Research Group: http://dsa-research.org
Globus GridWay Metascheduler: http://www.GridWay.org
OpenNebula Virtual Infrastructure Engine: http://www.OpenNebula.org