Now, i still don't know or fully understand all the intricacies of how the linux kernel works, but I think that most people would agree that, when computer lockups and failures happen, kernel modules and hardware drivers are usually to blame. So, here's the idea:

Build a layer of separation between the base Kernel, and the modules that are loaded into it, to enable better control how they access the kernel, and how the kernel should behave on the off-chance that one of these modules should break.

I think that something like this would help increase system stability and reliability by greatly reducing the number of system crashes, removing the need for doing hard-resets, which are two very serious issues which can break software, destroy personal data, and even damage hardware in some instances. I know that these things generally don't happen often, but the less they happen the better for everyone, and the more reliable our system becomes.

32-bit -server kernel or 64-bit -generic kernel ARE NOT A VIABLE OPTIONS.
This has been submitted many times, but people keep burrying it because of misinformation and plain ignorance.
Not having this thing makes Ubuntu worse for many.
Come on people, 4 GB RAM is becoming the standard.

32-bit -server kernel: unstable with nvidia/compiz, not configured for desktop, can open far fewer windows before start getting "black window" problem.

64-bit -generic kernel: on 4GB systems it actually results in having less usable memory than going with the standard -generic kernel which supports only 3GB.

Thanks for supporting this!

Hi folks!

Every time when I just want to remove/purge a old linux kernel, update-initramfs is called to recreate a initrd just before it will be removed from file system.

Take a look:

-----------
root@srvXY:/# dpkg -P linux-image-2.6.24-19-server linux-ubuntu-modules-2.6.24-19-server
(Reading database ... 30175 files and directories currently installed.)
Removing linux-ubuntu-modules-2.6.24-19-server ...
update-initramfs: Generating /boot/initrd.img-2.6.24-19-server
Purging configuration files for linux-ubuntu-modules-2.6.24-19-server ...
Removing linux-image-2.6.24-19-server ...
Running postrm hook script /sbin/update-grub.
Searching for GRUB installation directory ... found: /boot/grub
Searching for default file ... found: /boot/grub/default
Testing for an existing GRUB menu.lst file ... found: /boot/grub/menu.lst
Searching for splash image ... none found, skipping ...
Found Xen hypervisor 3.2, kernel: /vmlinuz-2.6.24-20-xen
Found Xen hypervisor 3.2, kernel: /vmlinuz-2.6.24-19-xen
Found Xen hypervisor 3.2, kernel: /vmlinuz-2.6.22-15-xen
Found kernel: /vmlinuz-2.6.24-20-server
Found kernel: /memtest86+.bin
Replacing config file /var/run/grub/menu.lst with new version
Updating /boot/grub/menu.lst ... done
-----------

Has all can see, update-initramfs recreate the file "/boot/initrd.img-2.6.24-19-server" but it's going to be removed on the next step!


[....]

The standard for linux sensors is lm-sensors. The best operating system for a linux server is Ubuntu. Fancier server-class machines which are now getting very cheap have IPMI for management. Without support for IPMI sensors, generally only CPU and HDD temp sensors can be read. For example, there are numerous Dell and IBM servers with IPMI now available on ebay and via other used outlets for around $200. Other distributions (e.g. RedHat) integrate this patch already.

http://bmcsensors-26.sourceforge.net/


Following is an illustrative example from my IBM eServer 325 (2xOpteron 246 - I paid $189 with 1GB ECC and 120GB IDE.)

$ sensors
k8temp-pci-00c3
Adapter: PCI adapter
Core0 Temp: +45.0°C

k8temp-pci-00cb
Adapter: PCI adapter
Core0 Temp: +40.0°C


$ sudo ipmitool sdr
Planar Temp 1 | 26 degrees C | ok
Planar Temp 2 | 29 degrees C | ok
CPU 2 Temp | 37 degrees C | ok
CPU 1 Temp | 41 degrees C | ok
Planar 12V | 12.47 Volts | ok
CPU 2 VCore | 1.55 Volts | ok
CPU 2 2.5V | 2.54 Volts | ok

[....]

I know that it would be hard but why don't offer support several UNIX kernels in the Ubuntu OS?

It would be awesome to join Ubuntu comunity with BSD and project Nexenta (http://www.nexenta.org/os) guys.

Later, when GNU/Hurd is completed it would be great to include that kernel as an option to the users.

I'm not talking about making more versions of Ubuntu, I'm talking about allow users to have installed several kernels in their machines and choose what they want to use.

NOTE: if you downvote the idea, please tell me why. Thanks.

when installing new hardware, or even a new laptop, or what not, one big show stopper is when some excentric piece of hardware locks the computer entirely (like what happens with my everex sa2053t notebook), this requires that you boot with a live cd with no extra modules being loaded, mounting the offending partition and blacklisting the module in question, this is something quite gruesome when you have to navigate a shady log and you end up wildguessing which is the offending module.

the idea is simple, during the boot process:

-check if module loading wasnt successful during previous boot. if it didnt load successfully, blacklist it.
-mark a module as being loaded.
-after loading the module, mark the module as successful.


this would not work with live cds, but it will be quite useful with a pendrive install stick, this would avoid many people needing the alternate boot disk with no gui.

Hello

As it seems hardy doesn't have support for it, because this path doesn't exist.

/proc/self/io

This feature is interesting for this little program called "IoTop".

http://guichaz.free.fr/iotop/

Regards

Bhavani Shankar.

Binary drivers on top of a stable application binary interface makes it feasible for hardware manufacturers to release device drivers without source code. Also drivers can be run with restricted rights.

You know, I ran into a kernel panic coming out of suspend the other day using Ubuntu's kernel. The only other time I've had a kernel panic was when I was using my own compiled kernel for an Operating Systems class.

In any case, considering that kernel panics are a very bad, and they're very hard for an ordinary user to diagnose, there should be a tool (ex. a message in the notification area the next time the user logs in) that easily allows a user to report them.

When there is a kernel upgrade for the system, I most often have to reinstall the openafs and virtualbox modules, since they don't get upgraded along with the kernel.

I would love it if the update manager could somehow keep track of manually added modules, and try to upgrade or recompile them (via module-assistant, maybe) as needed.