|Subject:||RE: Default Kernel .config file|
|From:||Harrington, Todd (Todd...@suntroncorp.com)|
|Date:||Jun 16, 2006 6:30:40 am|
-----Original Message----- From: Harrington, Todd Sent: Friday, June 16, 2006 9:10 AM To: 'Getting started with Red Hat Linux' Subject: RE: Default Kernel .config file
Thanks for your answer. It was very helpful and answered several smaller questions I had. I have built my new kernel, modules, modules_install, install. I then copied the /boot and the /lib directory back over to my boot server. I did not re-run mkinitrd because when I run the RHEL4 Network Booting Service (system-config-netboot) it asks me to point to a kernel and I point to /diskless/i386/RHEL4-ES/root/ (where it finds the new kernel that I just installed) and it then builds an initrd.img for me and sticks the initrd.img and new kernel into /tftpboot/linux-install/IMAGENAME There sizes look very good.
So my problem NOW is that when my diskless client boots the kernel loads and then I get this from initrd:
RAMDISK: Compress image found at block 0 - - Running /disklessrc Mounting /proc disklessrc: Line 71: /sbin/uniq: Permission denied disklessrc: Line 66: /sbin/sed: Permission denied disklessrc: Line 68: /sbin/uname: Permission denied disklessrc: Line 68: /sbin/grep: Permission denied (several more lines of this) Loaded nfs OK Mounting / disklessrc: line 163: /bin/cat: Permission denied Running dhclient bash-3.00#
At this shell all commands are permission denied.
This is so important that I get by this today. Do you know what my problem is?
-----Original Message----- From: redh...@redhat.com [mailto:redh...@redhat.com]On Behalf Of Rick Stevens Sent: Thursday, June 15, 2006 4:51 PM To: Getting started with Red Hat Linux Subject: Re: Default Kernel .config file
On Thu, 2006-06-15 at 15:08 -0500, Harrington, Todd wrote:
I am installing RHEL4 ES on a "client" computer. I do a fresh load to a local
hard drive on the "client" with the goal of eventually making this node a
"diskless client" that will boot via PXE.. I use the Netboot GUI on my Linux "server" (another machine) that comes with
RedHat and it pulls the image and kernel from my hard drive on my "client" and I
can now PXE boot my client from the server. I am actually now booting the
default kernel that was installed on the hard drive of my "client". Life is
My problem is that I want to reconfigure the kernel on my client and when I do,
the default .config files (in the "configs" directory or the Config-xxx file in
the /boot directory) do not contain the same settings as are in my original
kernel and my new kernel does not boot across the network. I have been pulling
my hair out trying to get my new kernel to boot. How can I get the default
.config settings for the kernel that was initially installed to my local hard
drive? I read about "extract_ikconfig" and /proc/config that allow you to pull
the .config info from a kernel but you have to have that support in the kernel
in the first place. It does not appear the initial RedHat install included this
in the kernel.
In the future, please format your messages as plain text, 72 characters per line. Not all mail readers deal with long lines well.
Now, to your problem...
When RHEL installs, the config file that matches the kernel that was installed is the "/boot/config-`uname -r`" file. The only difference between that config file and the actual kernel is that the kernel was built with the "CONFIG_DEBUG_INFO" option (in the "Kernel->Hacking" section) turned off, which creates modules that don't have full debug info in them and are therefore smaller in size.
Also note that the default kernels put all of the network, SCSI and non-ext2 filesystem drivers in the initrd image file. If you reconfigure the kernel, you have to rebuild the initrd image. That's done by:
# cd /boot # mkinitrd -f -v initrd-whatever.img whatever
replacing "whatever" with the kernel version number you're working with. Note that both the kernel AND the initrd image must be on your boot server or the kernel won't be able to find the drivers it needs.
Now, you CAN reconfigure the kernel and build the necessary network and filesystem drivers into the kernel and remove the need for an initrd image completely, but if you change the hardware (say, swap out an Intel e1000 card for an Intel e100 card), you won't be able to boot from the network again since the driver for the card won't be present.
My quicky document on building kernels and such can be found here:
---------------------------------------------------------------------- - Rick Stevens, Senior Systems Engineer rste...@vitalstream.com - - VitalStream, Inc. http://www.vitalstream.com - - - - The Theory of Rapitivity: E=MC Hammer - - -- Glenn Marcus (via TopFive.com) -
_______________________________________________ Redhat-install-list mailing list Redh...@redhat.com https://www.redhat.com/mailman/listinfo/redhat-install-list To Unsubscribe Go To ABOVE URL or send a message to: redh...@redhat.com Subject: unsubscribe
Hi, I am all set now. I was required to copy the file busybox.anaconda to my client's /sbin directory and when I did the permissions did not have execute permissions on it. I did a chmod to busybox.anaconda and now I can boot! Thank you to this mailing list.