Hi, I have the following problem: I'm trying to build a linux cluster of 16 processors (Pentium 4 530, 3.00 GHz). Each node has 2 Gb of RAM, 2 Hard Disks SATA (120 Gb each one), 1 CD-ROM driver, no floppy and a Gigabit Network Card integrated in the motherboard (master node also has a 100 Mbit ethernet card, not integrated). I installed Fedora Core 3 on the master and I installed Oscar 4.1. Everything were OK until the phase of Network Setup, after the creation of the image of the operating system (I applied the patch for Fedora Core 3 that I found on the home page of the Oscar website, following the instruction). OSCAR correctly collected MAC addresses of all the nodes; the problem is that when I network boot the nodes, they download the kernel boel and it appears that it can't receive other data from the master. Here there is the last part of the messages it writes on the screen:
Checking for floppy diskette. YOU MAY SEE SOME "wrong magic" ERRORS HERE, AND THAT IS NORMAL end_request: I/O error, dev 02:00 (floppy), sector 0 end_request: I/O error, dev 02:00 (floppy), sector 0 end_request: I/O error, dev 02:00 (floppy), sector 0 cramfs: wrong magic end_request: I/O error, dev 02:00 (floppy), sector 0 end_request: I/O error, dev 02:00 (floppy), sector 0 end_request: I/O error, dev 02:00 (floppy), sector 0 FAT: unable to read boot sector No floppy diskette in drive. read_kernel_append_parameters start_network IP Address not set with pre-boot settings. sleep 0: This is to give your switch (if you're using one) time to recognize your ethernet card before we try the network. Tip: You can use <ctrl> + <c> to pass the time (pun intended). dhclient Internet Software Consortium DHCP Client 2.0pl5 Copyright 1995, 1996, 1997, 1998, 1999 The Internet Software Consortium. All rights reserved. Please contribute if you find this software useful. For info, please visit http://www.isc.org/dhcp-contibut.html Listening on LPF/lo/<null> Sending on LPF/lo/<null> DHCPDISCOVER on lo to 255.255.255.255 port 67 interval 5 DHCPDISCOVER on lo to 255.255.255.255 port 67 interval 9 DHCPDISCOVER on lo to 255.255.255.255 port 67 interval 12 DHCPDISCOVER on lo to 255.255.255.255 port 67 interval 12 DHCPDISCOVER on lo to 255.255.255.255 port 67 interval 20 DHCPDISCOVER on lo to 255.255.255.255 port 67 interval 2 NO DHCPOFFERS received. No working leases in persistent database - sleeping. I couldn't configure the network interface using DHCP. Killing off running processes write_variables <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< This is the SystemImager autoinstall system Then there is the "Your autoinstall has failed" message and the prompt. From the prompt I saw that (apparently) there is no support for tg3 module of the kernel (or better, there are no modules at all), also because it seems to look for DHCP on loop ("DHCPDISCOVER on lo"). I tried to recompile the kernel and substitute it in the tftpboot directory. I downloaded the SystemImager 3.4.1 source code from sourceforge, I unzipped every tar.gz or .bz2 in the src directory and I tried to recompile the kernel 2.4.25 in that directory after applied all the patches, but the patches gave a lot of problems and building failed (I changed in each .patch file every reference to directory in order to let them match with the actual path of the source, but it still doesn't find some file that, in fact, doesn't exist). Anyway, I guess that tg3 support is native in boel, because I checked the linux.i386.config file and tg3 is in. So, what is the problem? I also tried to recompile the original kernel 2.4.25, but it gives the "unable to mount root fs on 01:00" error (that I found in the manual), so I can't avoid to use patches in SystemImager /src directory (that don't work). Could you suggest a solution? It is a problem of kernel boel that can't recognize my ethernet card? Or something else, for example that "cramfs: wrong magic" could indicate that there is some problem in mantaining the ram disk? There is an alternative way to copy the image to the nodes? For example, suppose I boot the node (one a time, obviously) with a Knoppix release. If I copy the image under oscarimage/ directory of the master on a USB device that I can mount under knoppix, it is possible to copy the image directly to the hard drives? How exactly is it possible to copy an operating system image? I guess a simple "cp" is not sufficient neither from master hard disk to USB and from USB to node hard disk. If this is possible, this "simple" copy would create the entire filesystem (that is, partitions, NFS home directory, etc...)? Thanks Lorenzo ------------------------------------------------------------------------- Visita http://domini.interfree.it, il sito di Interfree dove trovare soluzioni semplici e complete che soddisfano le tue esigenze in Internet, ecco due esempi di offerte: - Registrazione Dominio: un dominio con 1 MB di spazio disco + 2 caselle email a soli 18,59 euro - MioDominio: un dominio con 20 MB di spazio disco + 5 caselle email a soli 51,13 euro Vieni a trovarci! Lo Staff di Interfree ------------------------------------------------------------------------- ------------------------------------------------------- SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click _______________________________________________ Oscar-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/oscar-users
