|
On 06/08/14 21:45, Minalkumar Patel
wrote:
Today i reached migration of logical
volume using DRBD. I used ubuntu 12.04 with follwoing link:
From the URL you posted, it isn't very clear, but I imagine that you
will create an LV on both servers, then DRBD will connect the two
LV's together, and then you will use Xen/Remus to run the VM writing
to DRBD. I've been feeling most remus related discussion on the
list, due to interest, but not yet used it, so I'm not sure how it
handles both VM's writing to the same "disk". I assume DRBD proto D
has something to do with this, and if both machines write the same
data, then I guess it should be OK.
In any case, in my scenario, I use xen to run VM's, which talk to
iSCSI, which is actually two machines. Each san machine has LV's for
each VM, these are then "connected" with DRBD, and then the DRBD
volume is exported via iSCSI for xen. So regardless of which san is
currently the iscsi server, lvdisplay will always show all disks.
However, /proc/drbd will show which server is "Primary".
Again, with remus, it sounds like both servers will be Primary
though.
Perhaps a better suggestion would be to show the output that you are
seeing, and describe why you think it is wrong (if the above doesn't
help).
I'd be very interested to hear if you get this working, or from
anyone who is successfully using remus in production. There are some
VM's that are important enough where the overhead would be
reasonable.
Regards,
Adam
|
