Steve,
     Setting up NFS should be straightforward.  You install nfs-utils RPM,
set up an /etc/exports file, and do "service nfs start", that's it. Since
oscar set up the /etc/exports file for you, that step has been done.  So,
install the nfs-utils RPM, start the service, and try again, I think that
should do it.  Rich

Steve Muir <[EMAIL PROTECTED]> on 12/05/2001 04:04:55 PM

To:   Richard C Ferri/Poughkeepsie/IBM@IBMUS
cc:
Subject:  Re: [Oscar-users] newbie question



Hi,

Yes, I'm root when I do all this (I haven't set up another user account
yet).  I tried 'mount oscar:/tftpboot /mnt' and I still get a 'permission
denied' error.  I'm pretty sure I don't have NFS set up right.  I started
the service by running rpc.nfsd - I didn't edit any .conf files or anything
like I had to do to set up the dhcp daemon.  I'll go dig around for info on
setting up NFS properly.

Yes, all the machines are using the same Intel motherboard and the eepro100
seems to be the right driver.  Just for fun, I created the Etherboot
diskette using the eepro and tried booting and slave and it choked saying
it couldn't find the device.  The boot diskette I created with the eepro100
finds the device and starts booting the system (complaining that it can't
find any Stallion cards, 3c59x, etc.) and just hangs when it says it can't
find an NFS server/daemon.  I'll continue to give myself a crash course on
NFS.

Thanks,

Steve

--On Wednesday, December 05, 2001 3:54 PM -0500 Richard C Ferri
<[EMAIL PROTECTED]> wrote:

>
> Steve,
>      Let's go back to mount problem.  Were you root when you tried to
> mount the /tftpboot file system?  First, su to root and do the mount, as
> in:
>
> mount localhost:/tftpboot /mnt
>
> If that works, then NFS is happy and we can ignore NFS as an issue for
> now. If that fails, then NFS is unhappy and all the other stuff
> downstream is going to fail.   I've even seen weirdness where I boot 4
> nodes and NFS works on 3 of them, and 1 fails... when I reboot the
> failing node it works fine...
>
> The fact that you added a disk is a good thing, but we're not to the
point
> in the installation process where we look for the disk and format it.
So,
> the failure has nothing to do with having or not having a harddrive on
the
> comput node, yet.
>
> Did you say you were using eepro100 ethernet devices on your compute
> nodes? I'm sure we built eepro100 device support into the boot kernel,
> that's the most popular device.  However, as new versions of the device
> come out from different manufacturers, sometimes new device drivers come
> out and we have to rebuild the kernel.  So, if NFS is happy, I would say
> we have to rebuild the kernel with the latest device driver and see what
> happens...
>
> Rich
>
> Steve Muir <[EMAIL PROTECTED]> on 12/05/2001 12:45:24 PM
>
> To:   Richard C Ferri/Poughkeepsie/IBM@IBMUS
> cc:
> Subject:  Re: [Oscar-users] newbie question
>
>
>
> Hmmm...
>
> I added a drive to a client and I'm still getting the same error (though
> the Partition Check is displaying the client filesystems).  My
> /etc/exports looks thusly:
>
> /tftpboot      oscar001 (rw,no_root_squash)        #ADDED BY OSCAR
> /usr      oscar001 (ro,no_root_squash)        #ADDED BY OSCAR
> /home          oscar001 (rw,no_root_squash)        #ADDED BY OSCAR
>
> I've restarted the nfsd (which returns 'restarting NFS services
> rpc.mountd [ OK ]').  To answer your other questions, we're using RedHat
> 7.1 and Intel D850GB Pentium 4 motherboards with Intel on-board NICs.  It
> looks like the OS is OK with the driver/NIC combination - the interface
> on the server is up and when the client boots it seems to find the dhcpd
> on the server OK. Sorry, I'm not knowledgeable on NFS or I'd probably be
> able to figure this out myself.  Thanks for any help/ideas.
>
> Steve
>
> --On Monday, December 03, 2001 1:09 PM -0500 Richard C Ferri
> <[EMAIL PROTECTED]> wrote:
>
>>
>> Steve,
>>      That's correct, we do not do diskless clients.  However, the oscar
>> code is failing long before the disk gets involved. If you do not have
>> disks, then I would say oscar is not for you.  If you want to run
> diskless
>> for experimentation, we will be glad to talk to you off line and see if
> we
>> can help... Rich
>>
>> Steve Muir <[EMAIL PROTECTED]> on 12/03/2001 12:22:02 PM
>>
>> To:   Richard C Ferri/Poughkeepsie/IBM@IBMUS
>> cc:   [EMAIL PROTECTED]
>> Subject:  Re: [Oscar-users] newbie question
>>
>>
>>
>> I can't do the mount - I get a 'permission denied'.  Someone else from
> the
>> group said that OSCAR doesn't support diskless clients.  Does that sound
>> right to you?  If so, then I'll stopping beating my head against the
> wall.
>>
>> Thanks,
>>
>> Steve
>>
>> --On Monday, December 03, 2001 12:08 PM -0500 Richard C Ferri
>> <[EMAIL PROTECTED]> wrote:
>>
>>>
>>> There are two possibilites that I can think of: either NFS is broken on
>>> the server, or the kernel that you're network booting isn't happy about
>>> your eepro100 adapters.   Please post what your /etc/exports looks
like,
>>> and you might try mounting a file system from the server to the server,
>>> just to make sure NFS is really happy.    The mere fact that nfsd is
>>> running on the server has no bearing whether nfs is actually alive and
>>> happy.   Do something like:
>>>
>>> mount localhost:/tftpboot /mnt
>>>
>>> on the server and see if it mounts /tftpboot over /mnt.  If the mount
>>> fails, then it's an NFS problem, to correct do:
>>>
>>> service nfs restart
>>>
>>> and retry the mount above.
>>>
>>> The other possibility is that the kernel is somehow unhappy with your
>>> eepro100 adpaters on your client, and the kernel is failing to
configure
>>> them properly.  Since eepro100 is already built into the boot kernel
>>> (which kernel are you using to boot, by the way?) this is a much
nastier
>>> problem to debug.  Let's not think about rebuilding the boot kernel
> until
>>> we've ruled out nfs, ok?
>>>
>>> Rich
>>>
>>> Richard Ferri
>>> IBM Linux Technology Center
>>> [EMAIL PROTECTED]
>>> 845.433.7920
>>>
>>> Steve Muir <[EMAIL PROTECTED]>@lists.sourceforge.net on 12/03/2001
>>> 11:47:42 AM
>>>
>>> Sent by:  [EMAIL PROTECTED]
>>>
>>>
>>> To:   [EMAIL PROTECTED], [EMAIL PROTECTED]
>>> cc:
>>> Subject:  Re: [Oscar-users] newbie question
>>>
>>>
>>>
>>> Oh sure, it's network-booting OK.  DHCP is working and the client is
>>> booting off the server.  It looks like it's running some kind of
default
>>> script - it's looking for Stallion cards, then 3C59X, then CS89x0 then:
>>>
>>> IP-Config: No network devices available
>>> Partition check:
>>> Root-NFS: No NFS server available, giving up
>>> VFS: Unable to mount root fs via NFS, trying floppy
>>> VFS: Insert root floppy and press Enter
>>>
>>> No network devices available?  Huh?  Where?  I'm using eepro100 for the
>>> client's on-board NIC to get to this point.  Not sure what this refers
>> to.
>>>
>>> Steve
>>>
>>> --On Monday, December 03, 2001 11:19 AM -0500 Michael Chase-Salerno
>>> <[EMAIL PROTECTED]> wrote:
>>>
>>>> I sounds like the network boot wasn't sucessful at all. On the console
>> of
>>>> the  client, you should see it send out network requests and then load
>>>> the netboot  kernel and run it. Did any of this happen?
>>>>
>>>> Mike
>>>>
>>>> On Monday 03 December 2001 10:48, Steve Muir wrote:
>>>>> Hi anyone,
>>>>>
>>>>> I'm trying to set up an 8-node cluster - 1 server and 7 diskless
> slaves
>>>>> (using Etherboot floppies).  I've been through all the wizard steps
>> with
>>>>> no error messages and get to the point where it asks me to reboot my
>>>>> clients. So I watch as the machine boots and I get a "cannot mount
> root
>>>>> fs - trying floppy" and "NFS not running" or something very close to
>>>>> that and I'm stuck at that point.  I can see the client filesystems
>>>>> under /tftpboot/{client ip-address}/ and nfsd is running on the
> server.
>>>>> I'm guessing I need to explicitly export the filesystems but I'm real
>>>>> weak as far as nfs goes. Can anyone tell me if I'm overlooking
>> something
>>>>> really simple?
>>>>>
>>>>> Thanks,
>>>>>
>>>>> Steve Muir
>>>>> [EMAIL PROTECTED]
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Oscar-users mailing list
>>>>> [EMAIL PROTECTED]
>>>>> https://lists.sourceforge.net/lists/listinfo/oscar-users
>>>>
>>>> --
>>>> Michael Chase-Salerno       [EMAIL PROTECTED]
>>>> IBM Linux Systems Technology     Poughkeepsie, NY
>>>> System Installation Suite        www.sisuite.org
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Oscar-users mailing list
>>> [EMAIL PROTECTED]
>>> https://lists.sourceforge.net/lists/listinfo/oscar-users
>>>
>>>
>>>
>>
>>
>>
>>
>>
>>
>>
>
>
>
>
>
>
>








_______________________________________________
Oscar-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/oscar-users

Reply via email to