NFS connections irregular

Simon Slater pyevet at aapt.net.au
Thu Mar 22 14:17:18 UTC 2007


On Wed, 2007-03-21 at 19:01 -0600, David G. Miller wrote:
> Simon Slater <pyevet at aapt.net.au> wrote:
> > 	I have 3 computers A, B & C connected through an hub. B is a file
> > server running FC1, A & C have new installs of FC6. A & C can mount
> > shares on B.  However showmount -e gives:
> >
> > # mount clntudp_create: RPC: Program not registered
> >
> > from B to either of A or C; from C to A; A to C or itself.
> >
> > 	The mount -t nfs C:/share /mnt command from A returns:
> >
> > # mount: mount to NFS server 'C' failed: System Error: No route to host.
> >
> > similarly fron C to A.  Yet from server B the same command to either A
> > or C returns:
> >
> > # mount: RPC: Remote system error - no route to host.
> >
> > 	Each box can ping the others, the /etc/hosts and /etc/exports files
> > seem to be in order.  What am I missing?
> >
> > Regards 
> > Simon Slater
> NFS of FC1 vintage would probably still be using UDP.  Not sure if NFS 
> over TCP was even an option.  Check your firewall rules on B to see what 
> it does with UDP packets.  Ditto for A and C since B could be accepting 
> the packets but A and C throw them away at their firewall.  tracepath 
> (like traceroute but for UDP packets) may give you some additional 
> information.
> 
> It sounds like B has been around for a while and A and C are relatively 
> new installs.  Is this the case?  Did this just come up when A and C got 
> an FC6 install?
> 
> Cheers,
> Dave
	Here is the output of tracepath from box A:

[root at A ~]# tracepath B
 1:  A (192.168.0.3)                                     1.517ms pmtu
1500
 1:  B (192.168.0.4)                                    0.811ms reached
     Resume: pmtu 1500 hops 1 back 1 
[root at A ~]# tracepath C
 1:  A (192.168.0.3)                                     1.718ms pmtu
1500
 1:  C (192.168.0.5)                                   0.864ms !H
     Resume: pmtu 1500 
[root at A ~]# 

>From box B the first line is 1?:[LOCALHOST]... and the second line ends
with !H when pointed to A or C.  The tracepath output from C to B is
similar to the clip above from box A.  I'm not sure how helpful this is.
This has only been an issue since C came online a couple of weeks ago.
A & B are basically working as they were.

	Boxes B & C are both newly aquired (freinds were moving and found them
in the back shed) but all are PII's. B was setup to store A's files
while upgrading it from FC3 to FC6, but hard disk space was limited so I
did a minimal FC1 install as a temporary measure (that was 3 months
ago).  These 3 (and 4 more I've yet to look at) are destined for a LTSP
network spanning office and home.

Regards
Simon




More information about the fedora-list mailing list