🚀 go-pugleaf

Stephen M. Wynne (stevemw@wvirginia.wv.tek.com) wrote:
: We have a frequent problem with "hanging" on our system.  When use is
: up, users (provided they have NFS-mounted partitions in their paths)
: notice delays of 2 to 3 minutes (and longer) of commands as simple as
: "clear."

: We have 22 SGI systems at various stages of configuration/OS-level.
: The central file-server is a 4D/35 "Data Station" with two ethernet
: interfaces, one internal 1000MB and one external 1000MB disk, and
: running 4.0.1. Services provided include:

:   Euclid CAD database via NFS for about 15-20 heavy users at a time,
:   perhaps 30 or more at times (there are about 50 Euclid users).

:   PC-NFS to an unknown number of PC users -- probably about 50 at a
:   time or more.

:   NFS-mounted home, mail, X-terminal tftp (30 systems), and a common
:   directory. The /etc/rmtab contains 383 entries.

: Served include 21 SGIs including: 2 Crimsons, 4 Indigo R4000s, 2
: Personal IRISs, leaving the remaining Indigo R3000s. The X-terminals
: take XDMCP from the Crimsons and the R4000s for the most part.

: Shared on the network are 15-20 Sun workstations which have very
: little to do with the SGIs. The network rarely reaches 45% capacity.
: The Suns have few complaints about network performance.

: Intended changes include automouting home and work directories from
: the individual workstations across the SGI platforms.

: Thanks for any practial suggestions.

  Sorry to be so trivial, but have you tried to run more nfsd processes. I
mean like launching nfsd with nfsd 16 (in /etc/rc2.d/S??network).

: STEVE
--

-- Yaron.