Re: Windows Multi-node RAC (VMware)

Howard J. Rogers wrote:

> Yeah, but we're dealing with totally non-standard, non-certified
> implementations of some fairly exotic configurations here. And I guess
> we're at the mercy of VMware and how it does its stuff.

Granted.. but VMware is still utterly kewl. I use Vmware and then VNC as the desktop interface to the VMs. Geek candy. :-)

> No, 'fraid not. There's something about bridged networking which means
> that the two nodes don't communicate with each other properly.

Interesting. Have this problem been bounced off the VMware guys? They are a keen bunch.

> And I just want to clarify: it makes not a jot of difference to multi-Node
> RAC whether you specify fixed IP addresses or whether you make use of the
> cirtual DHCP server that VMware supplies.

Yep. On the clusters I worked with, we had a bunch of internal IPs (one per cluster node), and a single public IP for the cluster. We even used different IP classes (the cluster network in effect ran a private LAN). The public IP connections were routed (using a very basic load-balancing algorythm) to the least busiest node. Thus IP addresses as you say, is not an issue.

> It simply doesn't work with anything other than host-only networking.
> God knows why, but that's the way it is.

This is really strange...

> I don't think you need to be a member (but I could be wrong). It's just
> www.oaktable.net and then click the Files link.
>
> There are two of them: one on single-node-multi-instance-RAC (easy) and
> one on multi-node-single-instance-per-node-RAC (much, much harder).
> (incidentally, the single-node-multi-instance version works just fine and
> dandy with bridged networking. The multi-node version doesn't).

Thanks Howard, I will take a look and satisfy my curiosity. :-)

--
Billy