My original question:
>In the SunOS4.1.1 Release notes, chap 6 page 67 states:
>NFS clients now dynamically determine timeout values, by
>estimating the average and deviation of measured response
>times. When routing across networks, the transfer sizes
>are also adjusted automatically, to handle congestion better
>on wide area networks.
>My question is "will this feature work if the server is a non-sun machine,
>exporting to Sparc 2s running 4.1.1."
>From Barry Margolin <barmar@Think.COM>
Why shouldn't it? It's just adjusting the timeouts to match the actual
behavior, and adjusting transfer sizes when it appears that packets have
been dropped by gateways. What does the OS of the server have to do with
it?
>My concern wasn't so much with the OS as with the rpc.mountd daemon
>and the nfsd daemons on the server having something to do with this
>feature. I didn't think it would, but have been burned by this type of
>thinking before. We are going to mount over a wide-area net and
>I wanted to eliminate this thread of doubt.
>From Mike Raffety <miker@sbcoc.com>
I'd say all the relevant code is on the client side ... and I know the
"transfer size" can be set by a mount option, too, and therefore is set
on the client side.
>From stern@sunne.East.Sun.COM (Hal Stern - NE Area Systems Engineer)
this feature is entirely a client-side thing. the client
chooses its request size and timeout, and the dynamic retransmission
code simply modifies these defaults if the server is on another
IP network.
this may have some unwanted effects, like really poor write
performance. in general, i turn it off by setting "nfs_dynamic"
to 0 using adb.
>This removes the thread of doubt and adds something we may do
>on our next blockpoint for our local network desktops.
Andy Stefancik Internet: as6143@eerpf001.ca.boeing.com
Boeing Commercial Airplane G. UUCP: ...!uunet!bcstec!eerpf001!as6143
P.O. Box 3707 MS 64-25 Phone: (206) 234-3049
Seattle, WA 98124-2207
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:42 CDT