As I pointed out in a comment to the previous post I had got four instances up and running, but the install of the DB control software had failed on two of them. Tim suggested that I simply try recreating the repository on all four nodes (which might normally be considered drastic, but I didn’t have a working system, and I hadn’t done any customisation, so why not …). He pointed me at an article of his on trouble-shooting Enterprise Manager, and pointed out that I’d need to add the “-cluster” option. So I got the cluster and instances up and running (20 minutes to start it all up), and issued a single command:
emca -config dbcontrol db -repos recreate -cluster
Then I sat back and waited.
Twenty-five minutes later I had db control running on all four nodes, and I used Internet Explorer from the host machine to get the snapshot above.
Of course, now that I can do it, I probably won’t ever feel the need to do it again. I can’t get all four nodes running while my documentation VM (which is a Windows XP Pro) is running, so I’ll probably only run a maximum of three nodes at any one time, and often limit it to two. But it’s nice to know that I can do it if I really want to, and sometimes four nodes are necessary.
A couple more things that were causing me trouble which I’ve now fixed:
- I couldn’t get OEM from my documenation VM to talk to the RAC VMs – but it all worked properly (though slowly) after the re-install of the repository.
- I couldn’t get Samba to open Windows directories on the host (Windows 7) from the RAC VMs, though I could get to the other Windows VMs (XP Pro) with no problem – but mount -t cifs does what I really needed to do.