NCP timeouts

Oliver Xymoron (oxymoron@waste.org)
Tue, 5 Mar 1996 10:38:28 -0600 (CST)


I've got a box here running 1.3.68 and I'm having some difficulty with
volumes I've mounted with ncpfs. They will work for a while, but
occassionally the mount points will start showing empty directories. If I
unmount the volume (server) and then remount it, it works again for a
brief period. A look in /var/log/messages reveals:

Mar 5 09:41:53 death kernel: NCP server not responding, timed out
Mar 5 09:41:53 death kernel: NCP: notice message: result=-11
Mar 5 09:42:49 death kernel: NCP server not responding, timed out
Mar 5 09:42:49 death kernel: NCP: notice message: result=-11
Mar 5 09:42:49 death kernel: ncpfs: could not lookup vol SYS
Mar 5 09:44:04 death kernel: NCP server not responding, timed out
Mar 5 09:44:04 death kernel: NCP: notice message: result=-11
Mar 5 09:44:04 death kernel: ncpfs: could not lookup vol MFG
Mar 5 09:53:38 death kernel: IPX: Network number collision 5
Mar 5 09:53:38 death kernel: ^Ieth0 802.3 and eth0 802.2

The server in question is in another building which is connected by a
2Mb/s radio link. This link can occassionally get congested and cause up
to 5 or 10 second pauses for DOS/Windows clients, but it rarely causes
errors for them. The kernel timeout seems to be set to 6 seconds
(include/ncp_fs.h) - where does this number come from? DOS clients
(and other network file systems) appear to have timeouts of around a minute.
I'm going to experiment with changing this 60 seconds, any other advice?

--
 "Love the dolphins," she advised him. "Write by W.A.S.T.E.."