Well if you use CRS the 99.5% becomes cero no? :-) You could have asked him how do you upgrade the data dictinary without stopping.! Rgds LSC On 4/3/06, Michael Fontana <MFontana@xxxxxxxxx> wrote: > > > as well as the various comments about the 'rolling upgrades' via sql > apply > > that you have received, I'd point out that in fact GRID sometimes > requires > > more downtime than its single instance bretheren. For example from the > > Jan2006 CPU notes (for windows but the principle holds) > > > > > > "Special Installation Instruction for Customers using RAC for patching > > CRS: > > > > > > > > > > When patching CRS, please stop all 10g RAC instances and the Nodeapps: > > > > > > > > > srvctl stop database -d <DB_NAME> > > > > > srvctl stop asm -n <nodename1> > > > > > srvctl stop asm -n <nodename2> > > > > > srvctl stop nodeapps -n <nodename1> > > > > > srvctl stop nodeapps -n <nodename2> > > Niall, this is an excellent example of the point I was trying to make. > > In fact, the reason for my original post was that I had seen the > marketing presentation, and we had an Oracle customer engineer coming in > to do an additional presentation to management. I sprung some of the > stuff we've been discussing here, and got him to back off and say: > > In 99.5% of all cases, you can upgrade without downtime. > > I think that's still very much BS, but I didn't want to totally undress > him. > > At least I've put some doubt in my management's mind. > -- > //www.freelists.org/webpage/oracle-l > > >