1 person found this helpful
Shutting down a single RAC db instance, will not result in the diskgroup used by that RAC db to go offline - as there are still other RAC instances on other RAC servers using that diskgroup.
So if SAN/disk maintenance is to be performed, all RAC instances need to be shutdown and ASM used to confirm that the relevant diskgroup is no longer mounted.
Had a quick look and on 22.214.171.124 RAC (policy based), shutting down an instances does not auto dismount the diskgroup(s) used by that database instance on that RAC node. So you will need a manual dismount of that diskgroup on that instance - keeping in mind that the diskgroup will still be mounted on other servers by RAC instances using that database.
I'll try to explain the situation. We have a rac with two nodes, in two distinct spot. The diskgroup are stored in on the san. Now we must close the second spot for maintenance, and we must shutdown all the machine, because the electricity go down. my question is about the correctly sequence to stop all (instance,asm,diskgroup)
Shutdown o/s. On Unix/Linux, shutdown -h now. On Windows, shutdown /f /s (?).
If your RAC cannot survive a standard o/s shutdown, then you have serious h/w, driver, setup and configuration problems.
as I understand you will not shutdown any SAN or diskgroup, you only need to shutdown a node in your cluster.
If you do not want to do a hard shutdown. you can prefer to shutdown piece by piece and you can user srvctl for your operations just becareful to specify the node you want to shutdown on every srvctl command.
shutdown order will be
* grid infrastructure ( ASM instance)
then OS shutdown.
Why? Why an explicit and manual shutdown sequence?
Unless you have set up RAC using shoddy hardware, guess work for configuration, and a hefty dosage of ignorance?