Hi Amos,
I have done something like that with shared storage and your classic
linux heartbeat stuff a while ago. The setup I was referring to was
actually the storage devices for Xen nodes, so we are doing raid1 over
AoE (using vblade/LVM) across two whole machines. Kinda like a SAN AOE
storage device with no single points of failure. The project is not
quite finished yet, but should be as soon as I get some time. This sort
of setup requires the actually Xen Dom0s to be different machines,
bringing the machine count to at least 4. If you want just HA samba, you
only need two machines, even if you use DRBD to keep the disks in sync.
Samba is tcp though so if you want to keep the streams running you need
to use LVS, or something like the approach we are doing above, which is
Xen live migration for HA failover (this bit is not fully tested yet for
me).
One of the things I have learned the hard way is that clustering is very
much a word that can mean lots of things, kinda like content
management... so it all depends on what you are trying to do...
dave
Amos Shapira wrote:
On 10/01/07, David Kempe <[EMAIL PROTECTED]> wrote:
FWIW, I have setup an HA version of this recently and recent with some
hardware was giving me excellent throughput on Gigabit networks (haven't
By "HA" you mean that multiple hosts are connected to the same
disk-chain?
Care you give more pointers about such a setup?
Thanks,
--Amos
--
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
