[Pacemaker] why does pacemaker migrate a vm by stopping and starting instead of migrating action?

[mark - pacemaker list]

Hi Cherish,

On Wed, Dec 19, 2012 at 1:11 AM, bin chen <free2coder at gmail.com> wrote:

> Hi,all
>         My cluster is pacemaker 1.1.7 + corosync 2.0. I have write a
> resource agent to manage the virtual machine.The RA supports
> start,stop,migrate_from,migrate_to,monitor.
>         But when I try to migrate a running cluster vm(resource name is
> dcbh6f1c-GtNhnB-8597) from the host(h66)  to a target host (h65),the
> cluster just stops it in host A,and then start it in host B.
>


Can we safely assume that you are certain you have live migration working
without the cluster involved?  E.g. you can do an 'xm migrate' or 'virsh
migrate', whatever you're using to manage your VMs?

Have you set "allow-migrate=true" for the VM resource?  This snip is
showing the Xen RA, but the VirtualDomain RA has the same option, I believe:

---- 8< ----
primitive www3 ocf:heartbeat:Xen \
params xmfile="/etc/xen/www3" \
meta allow-migrate="true" target-role="Started" is-managed="true" \
---- 8< ----

If you already have that set, then your cluster configuration would be good
to see.

Regards,
Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20121219/3bcdfbf9/attachment-0003.html>