[comp.protocols.nfs] question on showmount and mountproc_dump

ginsburg@samsung.COM (Scott Ginsburg) (02/20/91)

I've written a simple showmount program to become familiar
with RPC/XDR and basic client programs and it works with one weird
exception.  I have one host on my system that produces the "can't
reply to host" syslog message when I direct my RPC mountproc_dump()
call at it using a UDP transport, which results in an RPC timeout
message on my client.  The program, however, works fine over a TCP
transport. How is this so?

						Thanks,
						Scott
-- 
------------------------------------------------------------------------- 
Scott Ginsburg WA2CJT	  Voice: 508-685-7200  FAX: 508-685-4940
Samsung Software America  Internet: ginsburg@samsung.com
Andover, MA

barmar@think.com (Barry Margolin) (02/20/91)

In article <19256@samsung.samsung.com> ginsburg@samsung.COM (Scott Ginsburg) writes:
>  I have one host on my system that produces the "can't
>reply to host" syslog message when I direct my RPC mountproc_dump()
>call at it using a UDP transport, which results in an RPC timeout
>message on my client.  The program, however, works fine over a TCP
>transport. How is this so?

Maybe that one host has a limit on the size of UDP datagrams that is too
small to hold its entire mount list.  TCP doesn't have any size limit on
message size.

I had a similar problem in an early version of Symbolics Lisp Machine NFS
Client, which only supported UDP transport, so had problems using servers
with large export lists.
--
Barry Margolin, Thinking Machines Corp.

barmar@think.com
{uunet,harvard}!think!barmar