Home / Troubleshooting / NFS Latency: Slow Reads, Writes, or Metadata

NFS Latency: Slow Reads, Writes, or Metadata

NFS mounts work, but users report slow opens, reads, writes, directory listings, or intermittent stalls.

What ONTAP is telling you

Latency is an end-to-end symptom. ONTAP workload latency and protocol connection data help separate storage/QoS pressure from network or client behavior; one average does not identify the bottleneck.

EMS reference No single EMS event is authoritative for NFS latency; correlate relevant NFS, network, disk, and hardware events.

How to confirm

qos statistics volume latency show -vserver <svm> -volume <volume>
qos statistics workload performance show -vserver <svm>
vserver nfs connection show -vserver <svm>
network interface show -vserver <svm> -fields lif,address,status-oper,curr-node,curr-port
event log show -time >1h -severity ERROR

Output shape: Latency output is broken into service-center components where supported; workload output includes operations and latency. Capture the same busy interval and compare affected with healthy workloads.

The fix

Act only on the dominant measured layer: remove or adjust an unintended QoS limit, address aggregate/node contention, or repair the network/client path. Preserve a before/after sample. A generic tuning change is not verified by the archive.

How to prevent it

Baseline latency and operations per workload, retain busy-window samples, review QoS assignments, and test the complete NFS path after network, export, or ONTAP changes.

Sources and version relevance

Commands and claims are limited to material already represented in the Black Box archive. CLI links target the archived ONTAP 9.19.1 reference; fields and availability can differ by release. Check your ONTAP version and on-cluster command ? before production use.

Related Black Box pages: NFS mount failures · performance triage · QoS · all troubleshooting guides

Related advisory: none identified in the archive for this operational symptom.