Fix NFS Client Input/Output Error on CentOS

This error even good Linux Admins make from time to time. An NFS client returns Input/output error (often as mount.nfs: Input/output error or I/O errors when reading an already mounted share) when the client cannot talk to the RPC/NFS stack it needs. Here’s the solution(s) that fixed it on CentOS, plus how to check the same pieces on modern RHEL-like systems.

What usually causes it

portmap (CentOS 5) or rpcbind (CentOS 6+) is stopped, so the client cannot resolve NFS-related RPC services.
The nfs-utils client tools and helpers are missing, so mount.nfs or the nfs kernel module helpers are incomplete.
Less often: the server exported the path with options the client rejects, a stale mount after a server reboot, or a firewall dropping RPC traffic.

Prerequisites

Root on the NFS client.
The NFS server hostname/IP and export path you are mounting.

1. Check that portmap / rpcbind is running

1) check that portmap start is running:

service portmap start
chkconfig portmap on

Once started try the mount again.

On CentOS 6 and 7 the service is named rpcbind instead of portmap:

service rpcbind start
chkconfig rpcbind on

or on CentOS 7 with systemd:

systemctl enable --now rpcbind

2. Install nfs-utils

2) install nfs-utils:

yum install nfs-utils

That’s it, you should be set, if one of those two does not work, then un-install and re-install nfs-utils:

yum remove nfs-utils
yum install nfs-utils

Then retry the mount:

mount -t nfs server:/export /mnt

Verify

rpcinfo -p localhost
showmount -e server
mount | grep nfs
df -h /mnt

rpcinfo should list portmapper/rpcbind. showmount -e should list the server’s exports (NFSv3). The mount line and df confirm the share is mounted and readable.

Troubleshooting

Still I/O errors on an existing mount: the server may have rebooted under a hard mount. Unmount with umount -f /mnt (or umount -l if it is stuck), confirm rpcbind and nfs-utils on the client, then mount again. See Troubleshoot Stuck and Hung NFS Mounts on AlmaLinux 10.
requested NFS version or transport protocol is not supported: different failure — the server is not offering the version/transport you asked for. See requested NFS version or transport protocol is not supported CentOS.
Firewall on the server: NFSv4 needs the nfs service (TCP 2049). NFSv3 also needs rpc-bind and mountd.

AlmaLinux, Rocky Linux and RHEL 8–10

Install the client package with dnf, enable rpcbind, and prefer an explicit NFS version:

dnf install nfs-utils
systemctl enable --now rpcbind
mount -t nfs -o vers=4.2 server:/export /mnt

portmap is gone on these releases; rpcbind replaces it. If you still use NFSv3, also enable rpc-statd and ensure the server’s mountd ports are reachable.

input/output error from NFS client on CentOS