Mark Do you mean the total failure part? Try it in a two node configuration, unplug the cable from lower node interconnect wait for 3 minutes and everything will be down. The reason is simple since the lower node will always survive Clusterware is forcing a node which has lost interconnect to survive, since the lower node does not have an interconnect Cache Fusion Fails, instance fails complaining about it. I have tested it over 20 times, I was puzzled at the beginning, I opened a SR to Support and they say that is how it works and I can file en ER if I want. Now, if you say you will use another network for Cache Fusion then it might work but then again Oracle recommends single network for Cache Fusion and Clusterware Network Heartbeat. Alex On 5/10/07, Mark W. Farnham <mwf@xxxxxxxx> wrote:
I have to disagree. If you have any link at all the disk holding the online redo logs of the non-surviving instance, then you can recover those logs and keep right on going. Even an nfs read only mount will do. It is of course also nice to be able to read the log and out directories of the non-surviving node if they are not totally shared as well, and if you need the other instance's archived redo for backup it must be accessible. Of course ideally all the disks are totally shared on bandwidth different from the cache fusion interconnect. Maybe you mean something entirely different from what I took your meaning to be… Regards, mwf ------------------------------ *From:* oracle-l-bounce@xxxxxxxxxxxxx [mailto: oracle-l-bounce@xxxxxxxxxxxxx] *On Behalf Of *amonte *Sent:* Thursday, May 10, 2007 4:31 PM *To:* K Gopalakrishnan *Cc:* Jeremy Paul Schneider; oracle-l@xxxxxxxxxxxxx *Subject:* Re: Voting disk TIE <snip> Two nodes is not really useful to check this since the lower node rule applies, which isnt quite good IMHO, losing interconnect equal RAC total failure Alex <snip>