I'm debating doing that. It's 16 boards, but guaranteed not to grow much beyond that. And one could symlink boffiles or something so you don't need to have 16 copies of that, too.
It seems as though the boot-up complaints about not being able to write to disk are relatively harmless. Tom On Wed, Jul 6, 2011 at 12:57 PM, John Ford <[email protected]> wrote: >> For those who care, you are just not supposed to enable eth0 in >> /etc/network/interfaces. I thought it wasn't showing up in ifconfig >> without doing that, but I must have been mistaken. >> >> Another question: what is the best way of making the NFS filesystem >> read-only so that the N ROACH boards do not conflict with one another? >> I modified fstab to have this line: >> >> rootfs / rootfs ro 0 0 >> >> There are a number of complaints along the way in bootup, however. >> Would it be better to edit /etc/rcSimple to remount read-only after >> boot? > > We install a different root file system for each roach. Not elegant, but > it works if you only have a few roaches to serve! > > John > > >> >> Tom >> >> On Tue, Jul 5, 2011 at 4:21 PM, Tom Downes <[email protected]> wrote: >>> I'm quite certain NFS works correctly since it's only the enabling of >>> eth0 that breaks things. >>> >>> I think the problem is described here: >>> >>> http://wiki.voyage.hk/nfs_voyage.txt >>> >>> What is happening is that /etc/init.d/mountnfs-bootclean.sh is >>> mounting "/" over NFS and then cleaning up after itself. It >>> inadvertently deletes a file important to DHCP, which sees the problem >>> and reacts by shutting down eth0 (and the NFS mount). >>> >>> I am going to remove the call to bootclean and see how it goes. >>> >>> Tom >>> >>> On Tue, Jul 5, 2011 at 4:16 PM, Adam Barta <[email protected]> wrote: >>>> Hi Tom >>>> >>>> >>>> Are you certain NFS is working correctly? Since the initial uImage >>>> transfer >>>> is over tftp and dhcp looks okay from the log! Possibly try mount nfs >>>> from a >>>> computer on the same subnet as the roaches using dhcp from the server? >>>> Maybe? >>>> >>>> >>>> Regards >>>> >>>> >>>> On Tue, Jul 5, 2011 at 11:56 PM, Tom Downes <[email protected]> >>>> wrote: >>>>> >>>>> Casper folks: >>>>> >>>>> I am setting up a netboot system for our ~16 ROACH boards. I had >>>>> things working until I attempted to edit /etc/network/interfaces to >>>>> enable eth0 to use DHCP. My testing to this point had been over the >>>>> serial port. As soon as the DHCP client makes a request, the NFS >>>>> connection breaks and I get: >>>>> >>>>> nfs: server 10.5.5.32 not responding, still trying >>>>> >>>>> This is, of course, after netboot makes some initial DHCP requests as >>>>> part of the BOOTP / TFTP dancing. All that works and it can boot >>>>> successfully if I do not enable eth0 as part of boot-up. >>>>> >>>>> How do I enable eth0 to use DHCP without breaking the NFS connection >>>>> that allows the ROACH to continue seeing the exported filesystem? >>>>> >>>>> Tom >>>>> >>>> >>>> >>>> >>>> -- >>>> Adam Barta >>>> c: +27 72 105 8611 >>>> e: [email protected] >>>> w: www.ska.ac.za >>>> p: about.me/adambarta >>>> >>>> >>> >> >> > > >
