Thanks to all who responded to my problem about nfsd and a very slow
3/470 server. Sorry it took so long to reply, I've been away for
a couple of weeks. Some people suggested that I change the number
of daemons from 8 to 12 or 16, which I did (in /etc/rc.local, change
'nfsd 8' to 'nfsd 16').
Others suggested that I use 'etherfind ' to figure out which
DL client (there are 11 on this server) is dragging me down.
The problem may have been that a client was executing a program
that has been deleted.
Sun suggested that I 'tune' my nfs by changing the mount parameters
in /etc/fstab on any machines mounting disks from the server in question.
I did this on a few machines.
The parameters I changed (added) were:
timeo=30, retrans=5, rsize=4096, wsize=4096
Example line from a clients fstab ( 'orion' is the 3/470 server):
orion:/home/orion/usr3 /home/orion/usr3 nfs rw,soft,bg,nosuid,timeo=30,retrans=5,rsize=4096,wsize=4096 0 0
( I know there is some debate over the use of soft mounts versus hard,
but the Sun rep suggested I use soft mounts in this case.)
Ultimately I could always reboot all the diskless clients and the server
and hope that the problem goes away.
The solution I found was to do a little of everything above. I added
a few more nfsd daemons, I 'tuned' my nfs and I also rebooted a
couple of machines. I havent seen the problem since.
Thanks to all who responded:
danaf@base.bellcore.com
kpc!cdr@uunet.UU.NET
butzer@cis.ohio-state.edu
micke@anl.luth.se
field@flash.cs.pitt.edu
stern@sunne.East.Sun.COM
mark@maui.Qualcomm.COM
macphed@dvinci.usask.ca
per@erix.ericsson.se
northrup@cps.msu.edu
rk@bartok.att.com
rlk@Think.COM
shj@ultra.com
Steve.Ackerman@zindigi.MSG.UVM.EDU
ssigv!lewis@uunet.UU.NET
oconnor!sbcoc.com!miker@oddjob.uchicago.edu
alastair@eucad.co.uk
dal@gcm.com
pren@BBN.COM
keves@meaddata.com
pbg@cs.brown.edu
picard!bill@att.att.com
phil@cs.arizona.edu
- wyllys Ingersoll
E-Systems, Inc
Melpar Division
-------
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:18 CDT