One or more NFS mount error(s) during SLES 9 boot.

(Last modified: 04Apr2006)

This document (10098671) is provided subject to the disclaimer at the end of this document.

fact

SUSE LINUX Enterprise Server 9

Gigabit switch, network, and cards

symptom

One or more NFS mount error(s) during SLES 9 boot.

mount server reported tcp not available, falling back to udp

mount: RPC: Remote system error

mount: backgrounding "10.1.13.3:/vol/dir1"

mount server reported tcp not available, falling back to udp

After boot, tests confirm that the mount has been successful and NFS over TCP is working fine.

After boot, stopping and restarting the NFS client (rcnfs restart) mounts without error.

cause

In some cases the local NIC driver may take a few seconds to fully initialize the card for communication.  Communication attempts made before the card is fully initialized can result in failure messages.

Switches can take some time to finish port scanning when a computer comes up.  This can vary from device to device and how they've implemented spanning tree technology.  Sometimes the better job the switch does in scanning the port, the longer it takes.  If the computer attempts communication (like mounting a file system over NFS) during that time, it can fail initially.

fix

The messages are effectively cosmetic, as it appears the system succeeds while retrying.  To help ensure initial success or to reduce the chance of the message occurring, try the following:

- Add a "sleep" delay in the /etc/init.d/nfs script, just before the "mount -at nfs" line (i.e.:  sleep 5).   The necessary length of the sleep delay (in seconds) may vary from one environment to another.
- Enable fast port scanning on the switch.  This can eliminate the errors or at least reduce the number of errors or the likelihood of occurance.
- To ensure repeated attempts to mount over TCP even after a failure, include in the fstab file nfs mount options 'tcp' and 'bg'.
- It is possible that configuring switch ports and NICs to know their speed or other settings (rather than auto-sense) would speed up the process and reduce or eliminate the chance of the error occurring.  (Not tested by the author).
- It is possible that completely disabling spanning tree port scanning would eliminate the error, as the switch ports would be put into forwarding state immediately.  However, this approach is ideal.  Spanning tree is implemented in order to prevent communication loops.

document

Document Title:	One or more NFS mount error(s) during SLES 9 boot.
Document ID:	10098671
Solution ID:	NOVL103176
Creation Date:	23Aug2005
Modified Date:	04Apr2006
Novell Product Class:	Linux

disclaimer

The Origin of this information may be internal or external to Novell. Novell makes all reasonable efforts to verify this information. However, the information provided in this document is for your information only. Novell makes no explicit or implied claims to the validity of this information.
Any trademarks referenced in this document are the property of their respective owners. Consult your product manuals for complete trademark information.