Re: [Kimchi-devel] Kimchi-devel Digest, Vol 23, Issue 121

On 10/27/2015 11:13 AM, Frank Novak wrote: > > > > > kimchi-devel-bounces(a)ovirt.org wrote on 10/27/2015 08:31:49 AM: > > > Date: Mon, 26 Oct 2015 16:08:11 -0200 > > From: <dhbarboza82(a)gmail.com&gt; > > To: Kimchi Devel <kimchi-devel(a)ovirt.org&gt; > > Subject: [Kimchi-devel] [PATCH 0/3] Live migration backend > > Message-ID: <1445882894-10208-1-git-send-email-dhbarboza82(a)gmail.com&gt; > > > > From: Daniel Henrique Barboza <dhbarboza82(a)gmail.com&gt; > > > > This patch series implements the first release of the Live (and

> > migration in Kimchi. > > > > *** Requirements for a VM migration *** > > > > - shared storage. The disk/storage path must be the same at the

> > and destination. > Is this a temporary restriction in kimchi? > We otherwise do support live migration w/ storage... > I guess I've expressed myself poorly there. What I meant was that the VM must be in the same storage pool at both hosts (shared storage). For example, a NFS/iSCSI storage pool that contains both the source and destination. The idea is that the migration process will only copy the VM RAM - the disks would be accessible by the VM at the destination from the same path. I am aware that there is an option to copy the storage (non-shared storage migration) but I've found out that it isn't that supported by Red Hat for example: https://access.redhat.com/solutions/60034 "Live migration with non-shared storage in RHEL6/RHEV3.* is a deprecated functionality, but under certain circunstances and with some limitations, it works fine. In RHEL7, this feature has been disabled." However, if some distro that Kimchi supports still have support for non shared migration, we can add this option as well.

> > > > - password-less login > > > > - enough resorces in the destination host to allocate the VM > > > > > > Limitations of this first version and possible candidates for future > work: > > > > - do not automate the process of password-less login (but it is able

> > verify this condition) > > > > - due to the above limitation, the 'user' parameter value is only

> > at this moment > > > > - do not verify the 'shared storage' condition > > > > - do not verify same hypervisor/arch conditions > > > > - the origin VM will shutdown after migration (not sure if can be > helped) > > > Ahh, shutdown where, if it was running on source, it should be running > on destination.. > clearly it will not be on source anymore.. > Yeah, that was my reasoning when seeing the VM shutting down automatically in the source after the migration was successful. Daniel