[mythtv-users] Questions on PXE booting a frontend

Jim Stichnoth stichnot at gmail.com
Sun Aug 2 03:38:19 UTC 2009


On Sat, Aug 1, 2009 at 7:45 PM, Paul Bender<pebender at san.rr.com> wrote:
> Jim Stichnoth wrote:
>>
>> I have a Zotac IONITX frontend that I'm interested in making
>> completely diskless and PXE booting, and I have a bunch of questions.
>> Currently it boots off a 4GB USB flash drive.  It is running MythDora
>> 10.21, using the kernel 2.6.27.9-159.fc10.i686.  I have rsync'd the
>> root file system to the backend machine, and that directory is
>> NFS-exported.  I understand that I will need to set up dhcp, tftp, and
>> syslinux (i.e. pxelinux.0) on the server.
>
> Why are you interested in network booting your frontend? Being the
> maintainer of MiniMyth as well as someone who network boots his MythTV
> frontends, I believe I have no prejudice against network booting MythTV
> frontends. However, while network booting has both advantages and
> disadvantages. Therefore, understanding your motivation for network booting
> would help with figuring out what combination of local and network software
> would be most appropriate.

I'm really just looking into this for the hobby aspect.  I love that
the ION allows a fairly low-cost, low-power, tiny-footprint, and
completely silent myth frontend.  Along those lines, I would like to
try to eliminate even the $15 flash drive, if possible.

I have two of these IONITX frontends, and I don't see having any more
in the foreseeable future, so I'm not doing this for system management
reasons.  I plan to just replicate the root file systems for the two
frontends, rather than the "pivot-root" approach described in
http://www.mythtv.org/wiki/Diskless_Frontend .

>> 2. Can the DHCP service in a home router and a Linux DHCP/PXE server
>> coexist on the same network?  I would prefer the home router to handle
>> most DHCP requests and the Linux server to handle just the PXE-related
>> requests for this frontend, so that e.g. my wife doesn't lose DHCP for
>> her laptop computer when the Linux server is down.
>
> In general, I do not believe that it is worth trying to run separate BOOTP
> and DHCP servers on the network. At best, it is a challenge to configure. At
> worst, the DHCP server does not support it.
>
> If you need to use an existing DHCP server, then you might consider booting
> using gPXE on a USB driver. You can hard wire it with the server (TFTP or
> HTTP) rather than depending on the BOOTP/DHCP server to provide it. However,
> it requires you compile gPXE in order to embed the server address in the
> gPXE binary. Again, whether or not this is the right solution depends on the
> reason for network booting.

Several people have already said this is unlikely to work, so the
consensus is probably right.  Here's why I was originally more
hopeful.  My home router's DHCP is already configured to assign static
IP addresses to all the myth boxes.  I thought to configure the Linux
DHCP server (if possible) to respond only to DHCP/PXE requests from
the two PXE-booting myth frontends, and to assign the same static IPs
as the home router.  With only two PXE frontends, that shouldn't be
hard to manage.  If both DHCP servers respond to a PXE request,
hopefully the frontend would ignore the response from the home router
since it would be missing the PXE extension.  At least that's my naive
hope from reading a bit about PXE.

>> 3. How reliably does NFS work as a root file system, if the NFS server
>> goes down and up?  I had trouble using NFS to serve up
>> videos/posters/pictures until I started using autofs.
>
> If the NFS server is not reliable, then either running the root file system
> from a local drive (whether magnetic or flash) or from RAM might be a better
> choise.

The NFS server in my case is the myth backend and it is pretty stable,
but once in a great while I have to bring it down for some reason.
When I started out, if that happened, NFS clients would give me the
"Stale NFS file handle" message until the NFS file system was
remounted.  I ended up putting in a cron job to detect this and
remount, until I learned about automount.  Maybe I was just using a
poor choice of NFS mount options.  But if the same thing happens to me
with an NFS-mounted root file system, I'm afraid I would have to
reboot all the frontends every time I did something to the backend.
We'll see, I guess.

Thanks to everyone so far for the advice.  I don't want to give up
quite so easily, so I'll see if I can add the right stuff to my initrd
image to achieve an NFS root for a normal non-PXE boot.  It seems like
I need that to be reliable anyway before adding the PXE part.

Jim


More information about the mythtv-users mailing list