Hi, I'm sorry but I haven't seen these problems before. Can anyone else on the list help with this?
One thing I will say is that there shouldn't be a scripts directory within /usr/share/systemimager/ (or at least there isn't on the clusters that I have setup this way). Also once you get far enough that the install scripts have run, depending on what action you told oscar to take upon completion, you may get an error. If you chose halt then you may get an error output when the script has completed. This can be fixed by changing the last line of the script to read 'shutdown' rather than 'shutdown -h now' (but can of course be ignored if you haven't chosen halt as your on completion action. Also you will need to add a line to the script just before the filesystems are unmounted as follows: chroot / echo 'MII_NOT_SUPPORTED="no"' >> /a/etc/sysconfig/network-scripts/ifcfg-eth0 (note that is one line) This will make sure that the e1000 network card starts properly on reboot within each of the nodes. Otherwise you'll get lots of errors because the network card has not come up properly when installation has finished. Sorry that I could not help more with the problem you are experiencing at the minute, but unfortunately I haven't got a cluster which I can play with at the moment to try and see what you are having trouble with. Hope you get somewhere soon, Rich ----------------------------------- Richard Bruin PhD Student Department of Earth Sciences University of Cambridge On Mon, 2004-06-07 at 02:55, pap4 wrote: > Hi Rich, > > Just a bit more progress. > I added "scripts" stanza pointing to "/var/lib/systemimager/scripts/" > into /etc/rsyncd.conf and the boot process went a bit further until it stopped > saying > > I will now try to get the autointall script: > rsync -aL 192.168.5.2::scripts/ /tmp/ > > write_variables > > run_autoinstall_script > > Please press enter to activate this console. > > I pressed enter and nothing happened. Well, there is no scripts write with the > names "write_variables" or "run_autoinstall_script" > > So, it looks like it get to the point where it is ready to copy the image of > the hard drive from the oscar server to the node... but it fails for some > reasons. > > Pavel. > > >===== Original Message From [EMAIL PROTECTED] ===== > >Hi there, > > > >Sorry I forgot to mention that for some reason even though there is a > >boel_binaries.tar.gz within the files that you downloaded, the boot disk > >still tells the machine to use the boel_binaries.tar.gz from the 'standard' > >flavour. > > > >To sort this you need to replace the file boel_binaries.tar.gz within the > >directory: /usr/share/systemimager/boot/i386/standard/ with the > >boel_binaries.tar.gz file which you have downloaded. This should fix the > >problem I think. > > > >Hope those instructions make sense, let me know if not. > > > >Rich > > > >----------------------------- > >Richard Bruin > >PhD Student > >Department of Earth Sciences > >University of Cambridge > > > >> Hi Rich, > >> > >> I downloaded the kernel and image files from the link below and used them > to > >> replace kernel and imigerd.img files in /tftpboot directory. BTW, I do not > >> have Systemimager installed on my machine, I have only latest version of > >> OSCAR. Ok..., after replacing those files the node boots, recognizes both > >> ethernet cards (e1000), gets an IP from the DHCP server,... and fails > during > >> executing command get_boel_binaries_tarball. It says > >> > >> rsybc -av server::boot/i386 ..... > >> failed to connect to 192.168.5.2: Connection refused > >> > >> By default Redhat does not use rsh anymore, only ssh. So, I guess the > >problem > >> is that the node simply can not get necessary files using rync since it > >> supposed to use ssh instead. > >> > >> Pavel. > >> > >> >===== Original Message From Richard Bruin <[EMAIL PROTECTED]> ===== > >> >Hi again, > >> > > >> >Here's a message I posted previously which has helped to get the intel > >> >e1000 cards working: > >> > > >> >http://sourceforge.net/mailarchive/message.php?msg_id=7530696 > >> > > >> >note though that the url listed within that message seems to have been > >> >cropped by the list software and should actually read: > >> > > >> > >>http://www.mail-archive.com/[EMAIL PROTECTED]/msg00350.htm > l > >> > > >> >I hope that helps, let me know if you need any more help. Hopefully > >> >these instructions should work ok for you though. > >> > > >> >Rich > >> > > >> >-------------------------- > >> >Richard Bruin > >> >PhD Student > >> >Department of Earth Sciences > >> >University of Cambridge > >> >Cambridge > >> > > >> > > >> >On Thu, 2004-06-03 at 16:00, Pavel A. Petukhov wrote: > >> >> Hi Rich, > >> >> > >> >> Yes, I am using standard boot disk flavour and yes I would like to try > >your > >> >> link. I did not try to test anything else yet except for this link > >> >> > http://www.revolutionlinux.com/downloads/OSCAR/Supermicro-P4SC8=P4SCi-en/ > >> >> > >> >> The problem with the link above is that those kernel and initrd.img > >are for > >> >> Mandrake. The remedy in this link works fine, the node boots, gets the > >> >> boot image from TFTP, gets to DHCP server, gets IP, boots, formats the > >> >> drive but fails to boot from the harddrive afterwards. This is something > >> >> that I expected since the kernel and initrd.img are for Mandrake. At > least > >> >> I know that the problem CAN be solved. So, now I have a choice either to > >> >> continue to try with RedHat or to switch to Mandrake. I still have a > >couple > >> >> of potential solutions suggested in the link below that I am going to > try > >> today > >> >> > >> > http://sourceforge.net/mailarchive/forum.php?thread_id=4831598&forum_id=1364 > >> >> > >> >> Pavel. > >> >> > >> >> -----Original Message----- > >> >> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] > >> >> Sent: Thursday, June 03, 2004 3:50 AM > >> >> To: Pavel A. Petukhov > >> >> Subject: Re: [Oscar-users] RE: DHCP Client failures after PXE Boot > >> >> > >> >> > >> >> Hi Pavel, > >> >> > >> >> Just a quick question here that might help with the installation > >> >> problem... Are you still using the standard boot disk flavour? If so, I > >> >> can send a link to a site which provides another flavour which I have > >> >> used in the past to get the e1000 network cards working. > >> >> > >> >> Hope this can help, just let me know if you haven't tried this yet, > >> >> > >> >> thanks > >> >> > >> >> Rich > >> >> > >> >> ------------------------------- > >> >> Richard Bruin > >> >> > >> >> PhD Student > >> >> Department of Earth Sciences > >> >> University of Cambridge > >> >> Cambridge > >> >> UK > >> >> > >> >> > We have exactly the same problem as Chad Dupuis. RedHat9.0 does not > >> >> > recognize IntelGigabit ethernet cards (e1000) in our Supermicro 3.06 > >> >> > Dual Xeon servers unless e1000-5.2.52 driver is installed. We can > >> >> > install new e1000 driver on the master node and bring up the network > >> >> > cards but we can not do the same trick with the computational nodes > >> >> > since OSCAR does not generate an image that has the correct e1000 > >> >> > driver built-in. As in Chad's case, the nodes PXE boot, then they get > >> >> > the boot image from TFTP server, and then try either to get DHCP > >> >> > address or to assign static ip and fail because there is simply no > >> >> > eth0 or eth1 interface. Exactly the same happens if we generate a boot > >> >> > disk instead of PXE boot. > >> >> > > >> >> > Can anybody suggest how to generate a RedHat9.0 boot image with the > >> >> > correct updated e1000 driver that can be used by OSCAR or how to force > >> >> > OSCAR to use the correct e1000 driver ? > >> >> > > >> >> > Pavel Petukhov > >> >> > University of Illinois at Chicago > >> >> > > >> >> > > >> >> > > >> >> > ------------------------------------------------------- > >> >> > This SF.Net email is sponsored by the new InstallShield X. > >> >> > >From Windows to Linux, servers to mobile, InstallShield X is the one > >> >> > installation-authoring solution that does it all. Learn more and > >> >> > evaluate today! http://www.installshield.com/Dev2Dev/0504 > >> >> > _______________________________________________ > >> >> > Oscar-users mailing list > >> >> > [EMAIL PROTECTED] > >> >> > https://lists.sourceforge.net/lists/listinfo/oscar-users > >> >> https://lists.sourceforge.net/lists/listinfo/oscar-users > >> >> > >> >> > >> >> --------------------------------------------- > >> >> Department of Earth Sciences > >> >> University of Cambridge > >> >> Downing Street > >> >> Cambridge CB2 3EQ > >> >> Phone: ( +44 ) 1223 333400 > >> >> Fax: ( +44 ) 1223 333450 > >> >> > >> > >> > >> > >> > >> ------------------------------------------------------- > >> This SF.Net email is sponsored by the new InstallShield X. > >> >From Windows to Linux, servers to mobile, InstallShield X is the one > >> installation-authoring solution that does it all. Learn more and > >> evaluate today! http://www.installshield.com/Dev2Dev/0504 > >> _______________________________________________ > >> Oscar-users mailing list > >> [EMAIL PROTECTED] > >> https://lists.sourceforge.net/lists/listinfo/oscar-users > > > > > >--------------------------------------------- > >Department of Earth Sciences > >University of Cambridge > >Downing Street > >Cambridge CB2 3EQ > >Phone: ( +44 ) 1223 333400 > >Fax: ( +44 ) 1223 333450 > > ------------------------------------------------------- This SF.Net email is sponsored by: GNOME Foundation Hackers Unite! GUADEC: The world's #1 Open Source Desktop Event. GNOME Users and Developers European Conference, 28-30th June in Norway http://2004/guadec.org _______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users
