WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-users

Re: [Xen-users] evms-ha + Xen

Steve Dobbelstein wrote:

If you use separate volumes for the VBDs for the domUs then no two domUs
will access the same area of the disk, regardless of whether the disk space
comes from a SAN or whether you are running Linux-HA or not.

You are correct but I need to be sure that the same domU cannot be started twice ...

The idea is to have 2 or more dom0s connected to common disk(s), I'm not a big fan of nfs root solution so we're looking at SAN.

Not being familiar with EVMS terminology yet so I'll revert to use of LVM terminology for now ...

If 2 dom0s try to bring up the same domU then we have a problem. I presume a cluster aware volume manager (evms-ha) would mark the LV as busy and there would be no fear of the LV ever being double mounted.

Also with evms-ha I should be able to create an LV from one node and have all other connected nodes immediately see it (to make live migration possible without manually rescanning other nodes for new LVs).


It would be helpful to know how you are using Xen and Linux-HA together.  I
can think of a couple different setups.  Are you running the domUs as
separate nodes in a cluster?  Are you running the whole physical machine as
a node in a cluster, for example, with another machine standing by as a
backup?  Some other setup?

Multiple physical nodes (dom0s) sharing some disk(s). I like the idea of evms-ha for the already mentioned (presumed) benefits - extra safety measures preventing split-brain (starting same domU in 2 places).

Heartbeat's exact role within the solution (if any) I haven't determined, mostly I'm looking at it because I believed it was required for evms-ha. Some other uses for heartbeat do come to mind of course:

* it would have a list of all domUs within cluster and possibly a preferred
  dom0 for each domU
* it could be used as a framework for migrating domUs (perhaps triggered by
  utilization, hardware issues, shutdown, etc.)

Thanks for the reply
--
Fraser Campbell <fraser@xxxxxxxxxx>                 http://www.wehave.net/
Georgetown, Ontario, Canada                               Debian GNU/Linux

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users

<Prev in Thread] Current Thread [Next in Thread>