but the vg can be accessed by the two host.
For example:
on xen1, there is vm1, vm2, vm3 and vm4 access from SAN by /dev/
domU_VG/vm{1,2,3,4}-disk
on xen2, there is vm5, vm6, vm7 and vm8 access from SAN by /dev/
domU_VG/vm{5,6,7,8}-disk
If xen1 comes down, vm1,vm2,vm3 and vm4 are reload (manually or by
script) on xen2
If there is intendance task on xen2, simply migrate vm{5,6,7,8} on xen1
There is no filesystem access simulty but the same vg is seen by the
two host.
thank's for your help ;)
Le 27 sept. 07 à 23:12, Chris Cox a écrit :
hubert chomette wrote:
the two machine use the same lun, on this lun, there is one vg with a
logical volume by virtual machine (10 virtual machine).
The idea is if one xen is down, the other can launch the virtual
machine
ad continu service.
Ok.. so the psedo HA model I mentioned. Access to the LUN isn't live
(in concept) but only becomes available to the failover system when
the primary system fails.
Nothing wrong with this. Remember, that when you disconnect
a LUN and reconnect it to a new host that it will scan what is
out there. The difficulty would be for hardcoded names of
volumes, etc, if you changed those things while the secondary
system is in charge (same issue with the config on the secondary
when the primary is in charge).
You might be able to do something
with auto discovery and a script rather than with fstab with
regards to bring LUN LVM filesystem back online. That's probably
how I'd do it... That way, you can at least take a good stab
of making sense of the LVM when it comes online.
Le 27 sept. 07 à 19:36, Chris Cox a écrit :
hubert chomette wrote:
Hi,
I recently install a SAN at my work. One lun is used for XEN and
can be
used by two xen server.
For more flexibility I use lvm on this lun. I was very happy of the
result, unless I discover (during a conversation with a friend)
that lvm
metadata can be compromised if the two server simulty modify
metadata.
- So what can I do?
Use seperate LUNs. I'm not sure what your friend is talking about.
Unless you have two machines using the same LUN, in which case
there are issues... this would be a cluster filesystem issue, or
near-high availability issue...
- Is it possible to allow LVM metadata modification by only one
server?
- should I use clvm instead?
Metadata won't creep outside of what is visible.
- Is it possible to migrate lvm to clvm without any data lost?
Not sure.
thank's for your help
regards,
ps: sorry for my poor english
_______________________________________________
linux-lvm mailing list
linux-lvm@redhat.com
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
--
Chris Cox
Sr. Unix Sys Admin
_______________________________________________
linux-lvm mailing list
linux-lvm@redhat.com
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
_______________________________________________
linux-lvm mailing list
linux-lvm@redhat.com
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
--
Chris Cox
Sr. Unix Sys Admin
_______________________________________________
linux-lvm mailing list
linux-lvm@redhat.com
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
_______________________________________________
linux-lvm mailing list
linux-lvm@redhat.com
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
