Why does NFS IO stop during takeover and giveback or during aggregate relocation
Applies to
- ONTAP 9
- NFSv4.x
- NFSv3
Answer
- I/O stops when a storage system serving data over NFSv4 to clients is in a takeover/giveback mode or when an operation such as an aggregate relocation is performed
- NFSv4.x is a stateful protocol similar to SMB/CIFS
- A short outage is expected anytime you change the interface you are connected to while using stateful protocols
- Applications are expected to recover during grace period
- When NFSv3 is being used, if an aggregate is relocated to another node, NLM lock reclamation occurs for the duration specified by
locking.grace_lease_secondsnode level option, therefore, the NFSv3 access will be affected as well.
Additional Information
- NFS recovery are client‑driven and depends on client TCP/RPC timeout and retry behavior.
- VMware ESXi may remain in APD state until NFS client timeouts expire and paths are re‑established.
- How to troubleshoot NFS APD issues on VMware ESXi
- Network File System (NFS) version 4 Protocol RFC 3530
- Specify the NFSv4 locking grace period (netapp.com)
- How does the NFSv4 grace periods work?
- Learn about adverse effects of session-oriented protocols during ONTAP upgrades
