|
Hi Carlos:
To make things simple, I
recommend you identify the Linux distro which will work with your hardware (from
the sounds of it, either Fedora Core 3 or Core 4). After that, re-install
the headnode with that OS and choose a version of OSCAR that supports it.
4.2.1b4 will be publically available shortly, it has support for both Fedora
Core 3 and Core 4.
This way, you eliminate
problems with having multiple installations of OSCAR and/or mixing Linux
distributions (future release of OSCAR will allow you to install a different OS
on the headnode and the computer nodes, but not right now).
Cheers,
Bernard
From: Carlos Jorge Gonçalves
[mailto:[EMAIL PROTECTED] Sent: Thu 02/02/2006 06:31 To: Bernard Li; [email protected] Subject: RE: [Oscar-users] Sata Disks and program parted Hello again Bernard,
After I read your message I tried to boot the new computer using the boot cd of FC2 and the I couldn’t recognized the hard disc!
So I the problem is really about FC2 doesn’t support my SATA disk L.
But now I have another problem. When I started installing my cluster I started width the FC3 but the version of OSCAR of that time (4.0) didn’t work well. So I made a downgrade for the operating system from FC3 to FC2 and everything went fine.
Now I have the following configuration
OSCAR is version 4.0 and is installed in /op/oscar-4.0 and I had a symbolic link (/opt/oscar pointing to /opt/oscar-4.0) that I use when I launch OSCAR The rpm files used to build the image are in directory /tftpboot/rpm-fc2-oscar-4.0 and I have a symbolic link (rpm) that points to the directory (rpm-fc2-oscar-4.0)
Suppose that:
The next time I started Oscar will the odd database (generated by oscar 4.0) be recognized by the new version?
Thanks, Carlos
From:
[EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Bernard Li
Hi Carlos:
Why don't you try using
the following newer version of SystemImager and see if it works (note this is an
unreleased version):
BTW, you probably want to make sure that Fedora Core 2 supports your hardware - I would recommend using at least Fedora Core 3 or 4 which will have better support for newer hardware.
Cheers,
Bernard
From:
[EMAIL PROTECTED] on behalf of Carlos Jorge
Gonçalves Hi,
I’m setting up a cluster using OSCAR (version 4.0).
Everything went fine and I succeeded in the installation of one cluster with 3 nodes (the main node where OSCAR is installed, and 2 client nodes).
The problems come up when I need to replace one of the nodes by another computer.
This new computer come up with a network card from Broadcom (model bcm57xxx 1/10/1000 Mbit/s).
The original files to support the PXEBOOT didn’t recognized this network card, so I download the drivers, download a new version of systemimager (version 3.6.2 that come up with kernel 2.6.12.2), and I also have downloaded and installed the Linux kernel 2.6.12.2.
I booted using kernel 2.6.12.2, compiled the network card drivers, and I copied the drivers (bcm5700.ko) into the directory initrd_source/my_modules (of systemimager).
Then I have compiled the new (first changed the FLAVOR file) kernel and initrd.img files, and then copied then to the /tftpboot/ directory and create a file with the MAC address of my network card (located in the directory /tftpboot/pxelinux.cfg/).
The new computer has SATA disks that where not compatible with the previous image that I had built for IDE disk.
So I create a new image (the new systemimager files had rename the mkautoinstallscript to si_mkautoinstallscript and mkrsyncd_conf to si_mkrsyncd_conf so I created symbolic links from si_mkautoinstallscript to mkautoinstallscript and from si_mkrsyncd_conf to mkrsyncd_conf)
Next I boot the new client, the client gets the correct files (kernel, initrd.img and boel_binaries.tar.gz) and the booting process continues.
Then, when the partitions are being created I get the following message for each of the partitions:
“The program parted is using a deprecated SCSI ioctl, please convert it to SC_IO”
The partitions are created and the process continues to end and the machine is ready to boot.
I reboot the machine but the boot from disk fails and the client node starts to download the image again.
I’m using:
I really appreciate help because I don’t known any more what I should do L Thanks in advanced, Carlos Gonçalves
|
- [Oscar-users] Sata Disks and program parted Carlos Jorge Gonçalves
- RE: [Oscar-users] Sata Disks and program part... Bernard Li
- RE: [Oscar-users] Sata Disks and program ... Carlos Jorge Gonçalves
- RE: [Oscar-users] Sata Disks and program part... Bernard Li
- RE: [Oscar-users] Sata Disks and program part... Bernard Li
