NFS

Most important commands to remember

• findmnt -t nfs,nfs4 — identify mounted NFS filesystems.
• nfsstat -m — inspect their negotiated mount settings.

Commands and flags

Command or option	Meaning
findmnt -t nfs,nfs4	Select mounts with either listed NFS filesystem type.
-o SOURCE,TARGET,FSTYPE,OPTIONS	Show server/export source, local mount point, type, and options.
nfsstat -m	Show information about mounted NFS filesystems, not a live throughput graph.

A source such as server:/export identifies a remote namespace. vers describes the NFS version; sec describes the authentication/security flavor. rsize and wsize, if shown, are transfer sizes in bytes, not measured transfer rates.

The concepts that matter

1. NFS makes remote files available through local paths

Network File System lets a client access files exported by a server. Once mounted, ordinary filesystem operations can travel over the network without an application constructing explicit network requests.

An export defines what the server exposes; a mount connects that remote namespace to a local path. Seeing a mount point does not mean the data is on a local disk or that the server is currently responding.

2. Identity must mean the same thing on both sides

The server applies file permissions to the identity carried by the request. With common AUTH_SYS authentication, clients supply numeric user and group identifiers. Mismatched IDs can therefore produce surprising access or ownership results.

Kerberos security flavors add cryptographic authentication, with integrity or privacy depending on the selected flavor. Export rules such as root squashing also affect identity handling. Local root authority is not automatically unrestricted authority on the server.

3. Caching changes when clients observe changes

NFS clients cache file data and metadata to avoid a network round trip for every operation. Consistency rules, open/close behavior, and version-specific mechanisms coordinate those caches.

Do not assume every read on every client is an immediate fresh server read. Nor should you assume all NFS versions behave identically. Applications sharing writable files must use access patterns and synchronization that match the filesystem’s guarantees.

4. Remote failure becomes filesystem waiting or errors

A server outage can make an application wait inside a file operation. That application may look stuck even though its own CPU use is low. The network and remote storage are now part of the file operation’s dependency chain.

Mount retry behavior matters. Hard mounts keep retrying certain failed operations; soft behavior can return errors and risks data corruption in some workloads. Changing that policy is not a harmless universal fix for a hanging application. Understand the workload before changing mount options.

One small example

Optional: inspect a Linux client with an existing NFS mount. This exercise does not mount a new export, change options, or read application files from the server.

findmnt -t nfs,nfs4 -o SOURCE,TARGET,FSTYPE,OPTIONS
nfsstat -m

Match each source and target between the two outputs. Look for the actual protocol version and security flavor. sec=sys does not provide cryptographic user authentication; it also does not tell you whether a separate transport-protection mechanism is configured.

If there are no NFS mounts, an empty result is expected, and findmnt may return nonzero. That is not evidence of a broken server. These local configuration views establish what is mounted and how it was configured, not current server reachability, latency, or cross-client consistency.

Keep this idea: NFS preserves the file interface while adding remote identity, caching, and network failure to every application’s storage assumptions.