Originally posted by adewri
but what about the raw files and the shared disk ?? does this solve the whole purpose of having a RAC... (High Availibility, Load Balancing)...
1. raw devices or files on any kind of a shared file system on shured disk arrays
ARE OPERATION SYSTEM STUFF,
this isn't oracle stuff.
but any way u can use raw devices with one rac instance,
this is not a problem.
2. HA and others topics of RAC are not subjects of this post.
-- question was "can we install only 1 instance of a rac "
-- anwer is "yes, we can"
Originally posted by adewri
And what next, how will i now add another node
This is not big deal too.
U have to:
-- copy whole $ORACLE_BASE to new node
(just because oracle installer makes the same work, using rcp)
-- correct some configuration files on new instance
then step by step:
1. install cluster software (os scpecific) + oracle custer software without cluster manager
2. start cluster software
3. start gsd on new node
5. manually activate new node, using srvctl utility and add new instance to this (rac) database.
(by insure u can relink oracle software after coping)
in database don't forget :
-- create addition redo tablespace for new instance
-- create addition redo log thread (+ files) for new instance
I'm curious. In TRUE sense is RAC Unbreakable ?? (May be True from Oracle Software perspective or from being safe from database crash on 1 node) Still we need SAN/ Sophisticated Storage mechanism to support Unbreakable database ? Right ?
After my expirence with :
RH Linux Advaced server + Oracle 9.2.0 + path (9.2.0.4) +
Oracle Shared filesystem i can say:
"Oracle RAC Unbreakable !!!" - is a little bit sacks.
But with RACs on Sun clusters wasn't a lot of problems.