NFS client becomes unresponsive (hangs) on SLES 11

This document (7008148) is provided subject to the disclaimer at the end of this document.

Environment

SUSE Linux Enterprise Server 11

SUSE Linux Enterprise Server 11 Service Pack 1

Situation

After a SLES 11 NFS client mount has been in use for some time, or under heavy load, the mounted area may stop responding (i.e. "hang").

This occurs when the NFS client system is noticing packet loss and attempts to recover by closing the NFS connection and opening another. (The user is usually not aware that any packet loss or recovery attempt has occurred.) The recovery process is flawed and can result in this failure. A LAN trace will typically show that while in this state, a new NFS client request will result in a new NFS connection being opened (a TCP SYN handshake to destination port 2049), followed immediately by the NFS Client closing the connection (a TCP FIN to port 2049) without requesting anything in between.

This is a client side problem, so it will typically effect only one NFS client at a time. If all NFS clients with a particular mount are unresponsive simultaneously, then that is likely a different issue, with something going wrong at the NFS Server side. However, it is conceivable that this bug could effect more than one client host at the same time by coincidence, i.e. if multiple systems were experiencing similar packet loss simultaneously.

Resolution

A fix has been checked into maintenance for SLES 11 SP1. It is available through Online Updates and other maintenance update channels, as kernel 2.6.32.36-0.5.2. This should be applied to any SLES 11 SP1 which is acting as an NFS client.

Disclaimer

This Support Knowledgebase provides a valuable tool for SUSE customers and parties interested in our products and solutions to acquire information, ideas and learn from one another. Materials are provided for informational, personal or non-commercial use within your organization and are presented "AS IS" WITHOUT WARRANTY OF ANY KIND.