How NVFAIL protects database files

The NVFAIL state is set in two cases, either when Data ONTAP detects NVRAM errors when booting up or when a MetroCluster switchover operation occurs. If no errors are detected at startup, the file service is started normally. However, if NVRAM errors are detected or the force-fail option was set and then there was a switchover, Data ONTAP stops database instances from responding.

When you enable the NVFAIL option, one of the following processes takes place during bootup.

If...	Then...
Data ONTAP detects no NVRAM errors	File service starts normally.
Data ONTAP detects NVRAM errors	Data ONTAP returns a stale file handle (`ESTALE`) error to NFS clients trying to access the database, causing the application to stop responding, crash, or shut down. Data ONTAP then sends an error message to the system console and log file. When the application restarts, files are available to CIFS clients, even if you have not verified that they are valid. For NFS clients, files remain inaccessible until you reset the in-nvfailed-state option on the affected volume.
Data ONTAP detects NVRAM errors on a volume that contains LUNs	LUNs in that volume are brought offline. Then the in-nvfailed-state option on the volume must be cleared, and the NVFAIL attribute on the LUNs must be cleared by bringing each LUN in the affected volume online. You can perform the steps to check the integrity of the LUNs and recover the LUN from Snapshot copy or backup as necessary. After all the LUNs in the volume are recovered, the in-nvfailed-state option on the affected volume is cleared.