toward the end of a class on friday, just for fun, i showed the students how to install git, clone the latest kernel source, and build and install a new kernel. since it was getting close to end of day, i wanted to keep it simple and directed them to just "make defconfig" to see what would happen.
the configuration and build of the kernel and modules worked fine, they installed the kernel and modules, then rebooted but the new kernel panicked almost immediately, apparently whining that it couldn't mount the root filesystem. since they were all running LVM-based systems, i assumed that perhaps we needed to explicitly add LVM support during the config step.
i'd still like to be able to build and boot a new kernel on an updated centos box so, later today, i'm going to reinstall 5.5 on a system without LVM (just define some fixed filesystems) to see if that makes a difference.
has anyone done this? should i be able to? to recap, all i want to do on a fully-updated centos 5.5 box is
* install the development tools * install git * clone the latest kernel repo * make defconfig * make * install new kernel and modules * reboot to new kernel
i'm willing to accept that it won't be a perfect kernel and perhaps some drivers will be missing, but at the very least, it should boot without panicking.
again, anyone else done this? should it work? i'll try this later today as soon as i have access to my test system again.
rday
On Sun, Oct 10, 2010 at 7:49 AM, Robert P. J. Day rpjday@crashcourse.ca wrote:
toward the end of a class on friday, just for fun, i showed the students how to install git, clone the latest kernel source, and build and install a new kernel. since it was getting close to end of day, i wanted to keep it simple and directed them to just "make defconfig" to see what would happen.
the configuration and build of the kernel and modules worked fine, they installed the kernel and modules, then rebooted but the new kernel panicked almost immediately, apparently whining that it couldn't mount the root filesystem. since they were all running LVM-based systems, i assumed that perhaps we needed to explicitly add LVM support during the config step.
(snip)
again, anyone else done this? should it work? i'll try this later today as soon as i have access to my test system again.
You can find the answer here :-)
http://blog.toracat.org/2010/03/want-a-custom-kernel-on-centos-noooooo-reall...
Akemi
On Sun, 10 Oct 2010, Akemi Yagi wrote:
You can find the answer here :-)
http://blog.toracat.org/2010/03/want-a-custom-kernel-on-centos-noooooo-reall...
that's it? enabling deprecated sysfs? cool, i'll check that later and report back, thanks.
rday
On 10/10/10 16:09, Akemi Yagi wrote:
You can find the answer here :-)
http://blog.toracat.org/2010/03/want-a-custom-kernel-on-centos-noooooo-reall...
Akemi
Quoted from the above link:
As always, Alan and Ned engaged their brains and we tried several things to find a solution. We were not going anywhere when vanecka posted a fix in this forum thread.
Just thought I'd mention, although not grammatically incorrect, the phrase is "going nowhere", as in "We were going nowhere when vanecka posted a fix in this forum thread.
Made me smile, but we all understand exactly what you mean :-)
On 10/10/10 17:05, Ned Slider wrote:
On 10/10/10 16:09, Akemi Yagi wrote:
You can find the answer here :-)
http://blog.toracat.org/2010/03/want-a-custom-kernel-on-centos-noooooo-reall...
Akemi
Quoted from the above link:
As always, Alan and Ned engaged their brains and we tried several things to find a solution. We were not going anywhere when vanecka posted a fix in this forum thread.
Just thought I'd mention, although not grammatically incorrect, the phrase is "going nowhere", as in "We were going nowhere when vanecka posted a fix in this forum thread.
Made me smile, but we all understand exactly what you mean :-)
Apologies - that was obviously intended to be off list :-)
On Sun, Oct 10, 2010 at 9:31 AM, Ned Slider ned@unixmail.co.uk wrote:
On 10/10/10 17:05, Ned Slider wrote:
On 10/10/10 16:09, Akemi Yagi wrote:
http://blog.toracat.org/2010/03/want-a-custom-kernel-on-centos-noooooo-reall...
Just thought I'd mention, although not grammatically incorrect, the phrase is "going nowhere", as in "We were going nowhere when vanecka posted a fix in this forum thread.
Made me smile, but we all understand exactly what you mean :-)
Apologies - that was obviously intended to be off list :-)
Corrections always appreciated.
What really amazed me is the fact you had not seen/decipher that blog for 7 months since I posted it back in March :-P
Akemi
On Sun, Oct 10, 2010 at 05:05:35PM +0100, Ned Slider wrote:
On 10/10/10 16:09, Akemi Yagi wrote:
As always, Alan and Ned engaged their brains and we tried several things to find a solution. We were not going anywhere when vanecka posted a fix in this forum thread.
Just thought I'd mention, although not grammatically incorrect, the phrase is "going nowhere", as in "We were going nowhere when vanecka posted a fix in this forum thread.
To continue off topic, I'd say that's debatable on this side of the Atlantic, though "getting anywhere" is probably more common.
Made me smile, but we all understand exactly what you mean :-)
You guys spell funny too, though I like that rock band from your country, the Roulling Stounes. (Yeah, I know I've used that joke privately, but what the heck, let me share my idiocy with the list.)
On Sun, 10 Oct 2010, Akemi Yagi wrote:
You can find the answer here :-)
http://blog.toracat.org/2010/03/want-a-custom-kernel-on-centos-noooooo-reall...
Akemi
and that did it, thanks. so now i'm curious -- do you know what specifically requires the deprecated sysfs format during boot? i followed your blog links and i didn't immediately see an explanation for *why* that fixed it. again, thanks.
rday
p.s. there were, unsurprisingly, a number of startup script failures but that's something i can handle later.
-
Akemi Yagi -
Ned Slider -
Robert P. J. Day -
Scott Robbins