Re: real applicaton cluster

> You've made a statement for which I can not find a single supporting
> document.

I call it bug, when the software doesn't have the same behavior as the document states. Oracle also calls it bug when they have a patch. Otherwise, they says it
is your problem- OS, configure problem ...

RAC on OCFS is my production box. I have spent so many time on this issue. I made three TARs. Oracle support does nothing, apart from asking for logs & upgrading to the next version of OCFS.

I have made two posts here.( many on linux group) http://groups.google.com.au/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&threadm=2d15bd69.0309251633.15ce77c2%40posting.google.com&rnum=1&prev=/groups%3Fq%3Docfs%2Bkernel%2Bpanic%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DUTF-8%26selm%3D2d15bd69.0309251633.15ce77c2%2540posting.google.com%26rnum%3D1

http://groups.google.com.au/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&threadm=2d15bd69.0309251633.15ce77c2%40posting.google.com&rnum=1&prev=/groups%3Fq%3Docfs%2Bkernel%2Bpanic%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DUTF-8%26selm%3D2d15bd69.0309251633.15ce77c2%2540posting.google.com%26rnum%3D1 Also have a look at 481888.995 in metalink. I still have such issue.

It's usable as long as you follow the rules.

1. Use linux file system for archived redo log. If the node dies and you need to see the redo log on the node for recover, manually mount it from the other node since it's on the shared disk.
2. Once the system runs for a while, don't access OCFS use os command. You shouldn't have such requirement, since you don't put any other files on that system.
3. Use RMAN for backup.
4. If you cannot create a new tablespace or add a new file from one node, try it from the other node. (Don't keep trying, even oracle cannot create a new file from that node. The other node may not have the problem at that time.)

Good luck,
Bin Received on Fri Mar 19 2004 - 23:38:46 CST