On Tue, Oct 03, 2000 at 06:10:52PM +0200, Manuel Bouyer wrote:
> Date: Tue, 3 Oct 2000 18:10:52 +0200
> From: Manuel Bouyer <bouyer@antioche.lip6.fr>
> To: Sam <sam@epita.fr>
> Cc: netbsd-users@netbsd.org
> Subject: Re: Nfs clients get frozen when NFS server crashes...
> X-Mailer: Mutt 1.0us
> 
> On Tue, Oct 03, 2000 at 06:03:29PM +0000, Sam wrote:
> > I know that. But we dont want to use a new major release, we wait til 1.5.1 or 1.5.2...
> > We had to much trouble with 1.4.0...
> 
> I have a NFS server running 1.5_ALPHA without problems:
> armandeche:/home/NetBSD-1.5/src/sys/net>rup disco
> disco                    up  82 days,   5:49,  load average: 2.39 2.72 2.81

You think NetBSD 1.5 is stable enough to serve 500 Pc 24h/24/h ? 
Because if it is, we will switch soon. :)

> 
> ALPHA ds20, 20 disks connected to 4 NCR 875 (raid 1), netgear gigabit
> interface, serving 15 solaris and 50 linux/windows machines.
> 
> At last it's better than FreeBSD requiring a daily reboot :)

Sure... But the FreeBSD were very stable when they served only 250-300 Pcs, i think that since 
our Computer Park got bigger we got more problemes.. It migh also be somone that floods the
servers, but we aint sure right now...

> 
> > 
> > > 
> > > > 
> > > > We have a problem now with the freeBSD servers, which nfsd process stops every day...
> > > > When this hapenns all the netbsd clients that try to access the exported file systems 
> > > > which are not accessible, get stuck, and sometimes get completely frozen.
> > > > 
> > > > Here is how they are mounted:
> > > > 
> > > > (sam@jotun 42)cat /etc/fstab
> > > > /dev/wd0a / ffs rw 1 1
> > > > /dev/wd0b none swap sw 0 0/kern /kern kernfs rw
> > > > /proc /proc procfs rw
> > > > hermes:/space/mail  /nfs/hermes/mail nfs nosuid,nodev,rw,soft,bg,intr 0 0
> > > > s1:/space/export    /nfs/s1/space    nfs nosuid,nodev,rw,soft,bg,intr 0 0
> > > > s2:/space/export    /nfs/s2/space    nfs nosuid,nodev,rw,soft,bg,intr 0 0
> > > > s3:/space/export    /nfs/s3/space    nfs nosuid,nodev,rw,soft,bg,intr 0 0
> > > > s4:/space/export    /nfs/s4/space    nfs nosuid,nodev,rw,soft,bg,intr 0 0
> > > > s5:/space/export    /nfs/s5/space    nfs nosuid,nodev,rw,soft,bg,intr 0 0
> > > > 
> > > > We then have to reboot the 500 clients each time the servers crashes...
> > > > It's really annoying and shouldn't happen.
> > > 
> > > Do the client use TCP or UDP mount ? Try to force UDP mounts.
> > > If this doesn't solve the problem, use tcpdump on the server after a reboot
> > > to see if the client tries to talk to the server, and if the server replies.
> > 
> > I dont know if it uses udp or tcp, i'll check that, is there an option to mount 
> > to tell wether to use UDP or TCP ?
> 
> 'man mount_nfs'. 
> 
> > I'll do the tcpdump stuff also..
> 
> You can also check with tcpdump if the client uses TCP or UDP ...
> 
> --
> Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
> --


----
sam@epita.fr
Admin - Sys 
EPITA (Ecole pour l'informatique et les techniques avancees)
Tel: 01 44 08 01 96