Can FlexCache running on clusters that are both 7-mode be upgraded to ONTAP 9?
Applies to
- FlexCache
- ONTAP 9
- 7-Mode
Answer
- Running FlexCache where the cache and origin clusters are running 7-mode and then upgrade of a single cluster creates a combination of 7-mode and ONTAP 9 has been seen to cause issues where upon upgrade the FlexCache relationship is broken and once recreated Write calls can fail silently due to protocol issues with the older NRV protocol used by 7-Mode FlexCache
- The main difference between the 7-mode features and the c-mode feature is the protocol that FlexCache uses communicating with the origin. 7-mode used NetApp Remote Volume(NRV) protocol running over the data ports. In c-mode the RAL protocol links the FlexCache volume to the origin
- Based on sktrace logging found in previous cases, the d-blade is expecting a cifs credential and returns the SPINNP_ERR_CRED_MISSING error to the n-blade. In the c-mode NFS protocol workflow, this error handled appropriately by fetching CIFS credentials from the credstore cache or sending RPC to SECD
- Based on code, NRV protocol has nothing provisioned for SPINNP_ERR_CRED_MISSING and it's mapped to the default "SERVERFAULT" error. We can see spinnp creds are populated with unix creds only and it looks like NRV is only working with Unix creds.
- Based on ASUP spinhi stats a jump for spinnp error 277(SPINNP_ERR_CRED_MISSING)in CREATE and WRITE calls should be noted
Additional Information
additionalInformation_text