ITEM: F4830L

TCPIP Configuration Problems involving HANFS

Question:

The customer does a NFS mount from a Native Netware Server (Novell) with
the NFS NLM loaded. The NFS client is one of the HANFS systems. This 
works fine when HANFS is down, but does not work when HANFS is up.

There are other systems on the other side of a bridge that cannot contact
one of the HANFS systems when HANFS is up.

Response:

Due to a separate problem,  the packet size on the Native Netware server
was set to 1490. The customer changed the MTU size of the network 
interfaces on the RISC System/6000s to 1482. NFS does not 
work well when the MTU sizes do not agree. The parameter name for the 
Netware server is the "maximum physical receive packet size".

The problem here is that HANFS on the primary server uses "ifconfig" to 
configure the primary network interface. This does not read any 
information from ODM, like the MTU size. When an interface is configured
using "ifconfig" is uses the default MTU size for a network interface.
It will not read a different MTU size value set in the ODM CuAt file.
The same problem will also occur on the backup during a takeover 
situation. The primary can be temporarily corrected by doing a 
"ifconfig en0 mtu 1490" after hanfs has completely started.

The different MTU size will impact takeover and reintegration in a
HANFS configuration (the different MTU size will not be used). 



Support Line: TCPIP Configuration Problems involving HANFS ITEM: F4830L

Dated: December 1993 Category: N/A

This HTML file was generated 99/06/24~13:30:51

Comments or suggestions?
Contact us