[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [Xen-users] Xen, Heartbeat2 and live migration

  • To: "Yann Cezard" <yann.cezard@xxxxxxxxxxx>, xen-users@xxxxxxxxxxxxxxxxxxx
  • From: "Petersson, Mats" <Mats.Petersson@xxxxxxx>
  • Date: Tue, 3 Jul 2007 11:10:10 +0200
  • Delivery-date: Tue, 03 Jul 2007 02:08:29 -0700
  • List-id: Xen user discussion <xen-users.lists.xensource.com>
  • Thread-index: Ace9TDsujITHOfVPTre6XBV9r1komgABNBxw
  • Thread-topic: [Xen-users] Xen, Heartbeat2 and live migration


> -----Original Message-----
> From: xen-users-bounces@xxxxxxxxxxxxxxxxxxx 
> [mailto:xen-users-bounces@xxxxxxxxxxxxxxxxxxx] On Behalf Of 
> Yann Cezard
> Sent: 03 July 2007 09:25
> To: xen-users@xxxxxxxxxxxxxxxxxxx
> Subject: [Xen-users] Xen, Heartbeat2 and live migration
> Hi,
> This question perhaps should be better posted on linux-ha but...
> it seems that Suse 10 SP1 allows to migrate Xen VMs between Heartbeat
> nodes using live migration, where other distros just stop 
> them on a node
> and start them on another node.

Whilst this MAY be true, I find it unlikely that there is any difference
between SuSE and other Linux distros. The key parts of Live Migration
(or any other migration) is done by Xen-supplied code. It is of course
possible that SuSE 10 SP1 has newer code than many of the other distros,
which of course would help the matter. 

Of course, if a node goes down in such a setup, the guests on that node
most likely can't be migrated, but have to be restarted from the
original config files - because if the machine has failed, it's not
likely to be "alive enough" to support migration. 

This of course also means that the "heartbeat" case is pretty much a
irrelevant factor [although of course, it depends on how you use
heartbeat - in the guest, or in dom0 (or both?). If it's in the guest, I
don't see why it wouldn't work, as long as you have a reasonable timeout
for heartbeat - the guest domain will be down for some dozen of
milliseconds or so - perhaps 100ms, so if you don't tolerate that sort
of downtime in heartbeat, then it's not going to work, and if you're
planning to run this sort of setup, it's probably good to make some
tests on how long the downtime is, and make sure that the "server is
dead" time about 2-3 times longer than the worst measured downtime for a
successful migration]. 


Xen-users mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.