On Sat, 5 Oct 2002, Brent Hasty wrote:

> On Saturday 05 October 2002 05:40, Stew Benedict wrote:
> > On Sat, 5 Oct 2002, Brent Hasty wrote:
> > > 2 issues
> >
> > 1 issue - your mailer borks the "Reply To:" :)
> >
> > > fist all my terminals are using the eepro100 ic with pxe.  how do I set
> > > the server up to boot the terminals using these nics with pxe?
> >
> > Sorry.  Several people have asked about PXE.  I know Erwan did some PXE
> > work for the clustering project, and I read the docs, but I don't have PXE
> > capable hardware to implement/test.  Patches welcome :)
> >
> > > second is when booting the terminals (using a boot flopy) it fails at
> > > mounting the root filesystem.  I have 5 partitions /, /usr, /var, /home,
> > > /tmp
> > >
> > > my exports lists the
> > >
> > > / (ro,all_squash)
> > > /home     192.168.1.0/255.255.255.0(rw, root_squash)
> > >
> > > do i need to modify my exports to reflect all system partitions?
> > > am i overlooking a step to give the terminals permission to mount the
> > > shared resources?
> >
> > That should be OK as you list.  I'm assuming the subnet is correct for you
> > network.  Any messages at the client or server end? Is clusternfs running
> > on the server?  You shouldn't have to mount all the partitions
> > specifically, it's only done for /home for rw access.
> >
> > Stew Benedict
> 
> well hopefully I will soon be retiring this old hp kyak terminal server based 
> on 8.2 and ltsp.  I will then be running under mdk9.0 and kde3 with all its 
> new goodies.  what do you mean by borks and how do I fix it?
> 

Your email doesn't have the cooker list "Reply To:" that most of the posts
do.  I guess it depends on your mail client.  Generally you just leave the
"Reply To:" setting alone.

> is it possiable to get the eepro100 to net boot off of the terminal server, 
> as though using a boot rom rather than pxe?
> 

If you're getting to the point of nfs mounting, it sounds like you are
able to boot the eepro100, or you're using a different card for that?
I had one 3com card here that would do the initial boot, but then would
fail asking for the IP a second time.  Erwan told me that it was because
it was trying to talk to a pxe server, which didn't exist. (I may not have
the terminology exactly right here).

> I will see about the possiability of patches, but it would be a learning 
> project for me to do.
> 
> how do i check the status of cluseter-nfs?
> 

It's started by a normal initscript, just like nfs-server.  Should be 2
processes running:

[root@larry root]# service clusternfs start
Starting ClusterNFS (rpc.mountd):                               [  OK  ]
Starting ClusterNFS (rpc.nfsd):                                 [  OK  ]

On occasion rpc.nfsd doesn't start the first try.

Stew Benedict

-- 
MandrakeSoft    
PPC FAQ: http://www.linux-mandrake.com/en/ppcFAQ.php3
IRC: irc.openproject.net #cooker-ppc


Reply via email to