both the ACFS Registry and the Single File System Resource are managed under CRS. Just different ways of going about it.
If you don't need another resource depending on this file system, then either one works..... :)
In the end, both are provided so that you can choose which management interface makes more sense to you!
An update. ACFS now comes up after reboot, but it was far from straightforward. Testing this without a complete reboot e.g. 'just restarting the clusterware' is not an adequate test -- ACFS will come up mounted upon clusterware restart, but not on a reboot if the registry entry doesn't exist. see tech note 1486208.1.
+“ora.drivers.acfs” Resource Was Not Configured Therefore RAC ACFS Filesystem Is Not Mounting During The Reboot. [ID 1486208.1]+
The solution was acfsroot install / enable. Check the clusterware stack for an entry of 'ora.registry.acfs' afterwards. ACFS should now return after reboot.
Here's my question -- why isn't this being installed after a clean grid installation? From what I'm reading (see below), it sounds like root.sh should handle this. why would Oracle need a tech note on this issue if this were being handled as advertised? I just installed 3 clusters and all of them had this issue, missing ora.registry.acfs entry. All of them had root.sh & orainstRoot.sh run cleanly. This is OEL6 / 18.104.22.168.
"The Oracle ACFS drivers resource (ora.drivers.acfs) is *created by the Grid Infrastructure root script that is executed following the Grid Infrastructure installation*. The Oracle ASM instance resource (ora.asm) names the drivers resource as a weak dependency. As a result, the start action for the drivers resource is also called whenever the start action for the ora.asm resource is issued. The start action for the drivers resource includes support for loading the Oracle ACFS, Oracle ADVM, and OKS drivers into the operating system. "
@user9308584 : Thanks a lot!
I was having this same problem, when I was bringing CRS down, it was looping to unmount ACFS disks and would never do it as long as users were connected there.
To fix it I had to register the disks with srvctl instead of acfsutil.
Now CRS goes down without any problems.