Hello, We have a few old PCs ( PentiumI's and Celerons) which I was trying to install Oscar on. The server is running Red Hat 7.3. The server as well as the clients all have 3Com Etherlink III network cards ( ISA). We have exactly the same problem as the one described below. When network booting clients, the clients seem not to detect their own network cards and broadcast only on loopback interface. The support for 3Com cards is added in /usr/share/systemimager/i386-boot/config-standard-2.1.3oscar from the start. Now we recompiled the kernel, but I am not quite clear on what you meant by it. My understanding is that when the autoinstall diskette is created the kernel and initrd files that are used to create it are: /usr/share/systemimager/i386-boot/kernel-2.1.3oscar and /usr/share/systemimager/i386-boot/initrd-2.1.3oscar
How can I create a new version of kernel-2.1.3oscar ? Recompiling the kernel and stripping it to fit onto autoinstall floppy is probably not the best idea. I would really appreciate some advice on what to do. Thanks, Daria On 15 Aug 2002, Michael Chase-Salerno wrote: > It looks like another case of the kernel shipped with OSCAR not > supporting your network card. Its the same in either case because both > PXE and floppy boot both end up running the same kernel, its just a > matter of how they get it. Did you try the other kernel from the FAQ > item that I believe Jason pointed out? That may work for you. If not, > you can build a kernel that has your support compiled in. Since you are > PXE booting, you don't have to be concerned with fitting it on a floppy, > so you should just be able to take the .config file from SystemImager > (/usr/share/systemimager/i386-boot/config-standard-2.1.3oscar) and add > your support in. Then build a new kernel and replace the one in > /tftpboot with it. You could also try the standard RedHat kernel from > the server, but I'm not sure if that would work or not, but it would be > quick to try. > > Its normal for the client to attempt to broadcast over lo0, the problem > is that its not trying eth0 also because it doesn't recognize the card. > > The ip_assignment method is really up to you, I'd avoid static_dhcp as > it does some things that aren't what you might normally expect. So > either use dynamic_dhcp if you want the clients to get their IPs from > the server or static if you just want them set at install time. > > Mike > > On Thu, 2002-08-15 at 10:03, Bruce Becker wrote: > > Hi guys > > > > So, we have OSCAR-1.3 on Linux RedHat-7.2, with D-LINK DFE-550TX NIC's > > which are PXE-2.0 compatible and a CHRONOS switching hub and we are trying > > to build a cluster. All steps up to and including step 4 go off without a > > hitch. We have as of now NOT been able to network boot our clients, however ! > > > > I think I have tried everything - network booting using PXE does not work > > and network booting with a floppy has given us the same error as the > > former method. RH-7.2 DOES support our cards. We are now very stuck as to > > what to do, since I think we have exhausted all options and are obviously > > missing something. > > > > Included is our previously posted message (see below). What confuses me is > > that when it is booting, the machine seems to be listening for the DHCP > > broadcast on the loopback device lo, instead of the NIC, eth0. Surely this > > is not right ? Or is it ? If not, how can we change it ? > > > > When we build the image, is it correct to select the dynamic_dhcp option > > as our IP assignment method, or should we choose static_dhcp, or static ? > > At the moment, we have a standalone toy cluster of 2 clients and a > > server... > > > > We are getting quite desperate at this stage, so any insight would be > > greatly appreciated > > > > > > -------------------------------------------------------------------- > > Hi Mike, thanks for answering so soon. I read your mail and decided to > > ./start_over > > > > Now, we get the same error, but let me be more explicit. > > Again, we are using RedHat-7.2 and OSCAR-1.3 > > We have PXE-compatible cards and we are using PXE to boot the clients. PXE > > works fine, I think, since we get the PXE menu option and Linux's kudzu > > had no problem identitying the cards. So, I am assuming now that the > > problem is not hardware-based. > > > > Let me tell you what we did. > > After ./start_over, we once again followed the step-by-step instruction > > manual. > > > > Step 1 went off without a hitch. > > Step 2 (build SIS image) : I chose the standard rpmlist and partition > > files and set the IP assignment method to dynamic DHCP, asking the > > machines to shutdown after install. that all worked fine > > Step 3 : we have at present two nodes, so they were defined as > > oscarnode1.cluster and oscarnode2.cluster and given IP adresses of > > 192.168.1.2 and 191.168.1.3, because the server is 192.168.1.1. No errors > > were reported > > Step 4 : MAC adresses were seen, collected and assigned to nodes, no > > problems. After collecting MAC adresses, we configured the DHCP server (ok) and >then > > rebooted > > > > Now, is where the problems started > > > > When the clients were booted, they correctly booted from the NIC (because > > all other options were turned off in the BIOS) and saw a kernel and then > > looked like they were happily booting. The first wierd thing they did > > was try to look for a .cfg file in /floppy (even though we have no > > floppy drives). Then, they stared listening on the loopback device and > > (obviously ?) didn't hear anything. Then, another thing I thought wasn't > > right was that it had the wrong netmask (I told it 255.255.255.0) - we got > > this message: > > ================================================ > > DHCPDISCOVER on lo to 255.255.255.255 port 67 > > interval 8 (and some other intervals) > > NO DHCPOFFERS RECEIVED > > No working leases in persistent database > > I couldn't configure the network interface using DHCP > > Your auto-install has failed > > =============================================== > > > > When we rebooted the machines, there was nothing installed, which was not > > a surprise. So, we could not proceed to step 5. > > > > What have I left out here ? > > thanks in advance, > > > > -- > > Bruce Becker, PhD student - Department of Physics > > University of Cape Town > > > > Room 405, R.W. James Building, UCT > > University Avenue North > > Private Bag RONDEBOSCH > > 7700 > > > > tel (w) +27 21 650 3356 > > tel (m) +27 82 537 9425 > > fax +27 21 650 3342 > > > > http://qgp3.phy.uct.ac.za/index_becker.html > > > > > > > > > > > > > > ------------------------------------------------------- > > This sf.net email is sponsored by: OSDN - Tired of that same old > > cell phone? Get a new here for FREE! > > https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390 > > _______________________________________________ > > Oscar-users mailing list > > [EMAIL PROTECTED] > > https://lists.sourceforge.net/lists/listinfo/oscar-users > > > > > ------------------------------------------------------- > This sf.net email is sponsored by: OSDN - Tired of that same old > cell phone? Get a new here for FREE! > https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390 > _______________________________________________ > Oscar-users mailing list > [EMAIL PROTECTED] > https://lists.sourceforge.net/lists/listinfo/oscar-users > ------------------------------------------------------- In remembrance www.osdn.com/911/ _______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users
