So I wanted to start playing around with Docker and LXC containers and all kinds of other fun stuff on one of my cheap Kimsufi dedicated servers. I immediately ran into a problem though, Kimsufi (and parent company OVH, and probably sister company SoYouStart) use a modified Ubuntu Kernel. Why do they do this?

In short: optimization, speed, compatibility, control, etc. You can read a bit more about it here if you care to:

That’s fine for 90% of what most people do, not so much for virtualization and quasi-virtualization type stuff like Docker and LXC containers.

I ran into a problem on the first step when I went to install the kernel extras package. It looked something like this:

blah@blah:~# apt-get install linux-image-extra-`uname -r`
Reading package lists... Done
Building dependency tree
Reading state information... Done
E: Unable to locate package linux-image-extra-3.14.32-xxxx-grs-ipv6-64
E: Couldn't find any package by regex 'linux-image-extra-3.14.32-xxxx-grs-ipv6-64'

Because we are running a custom kernel, we can’t get (or at least I wouldn’t go down the road of trying…) a compatible linux-image-extra package (or it is something included with the generic kernel and not with the OVH custom kernel, I am not very clear on this point).

So I started searching and came across some guides on reverting to a standard kernel…

First, I ran “uname -r” to get my current kernel:

blah@blah:~# uname -r

That is showing the custom kernel. After the kernel swap was all said and done I ended up with:

blah@blah:~# uname -r

Which is a vanilla, stock, ubuntu kernel.

How I got there was by following this guide (I had to translate from french, thanks google…):

To summarize, here were the commands/steps:

sudo su
apt-get install linux-image-server
mkdir ~/ovh.d
mv /etc/grub.d/06_OVHkernel ~/ovh.d/

This worked great for me, however, a few words of caution… You should make sure you have any important files backed up off-system before you start this as you could easily end up with a system that isn’t easy to boot or work on, especially remotely. After you run the command to move the existing OVHkernel file, you should ls in the /etc/grub.d directory to make sure it is gone and you should also ls in the ~/ovh.d directory to make sure it was moved to the right place. Be sure to run “update-grub” before rebooting.

I was surprised at how easy it was to make the switch. Thus far, no adverse affects although I am sure my server is no longer as lean as it could be.


1 of 1

13 comments on: Switch to Standard Ubuntu Kernel on OVH or Kimsufi Servers

  1. Chris

    cheers, wasted a lot of time before I found your post.

    • nbeam

      LOL – thank you… I also wasted a lot of time before I found the referenced posts in my article 🙂 – I just got a new beefier server from SoYouStart (another daughter company of OVH, like KimSufi) and plan on diving into open-source virtualization next so figuring out how to switch to a stock kernel was helpful.

  2. Sam Howell

    Oh my gosh, thanks for that. Took me ages to figure out how and then I found this article!

    • nbeam

      LOL – No kidding – took me a lot of digging. Glad this was helpful.

  3. Sam Howell

    Hey quick question, how would I go about restoring the OVH kernel?
    I would imangine I would move the folder in ~/ovh.d back to /etc/grub.d/ but what do I do after that? remove all the files from /etc/grub.d/?

  4. rob cousins

    thankyou this worked for, god know why that are runinng kernal 3 when using ubuntu 16.

  5. rob cousins

    sorry typo..thankyou this worked for me 🙂

  6. Chris McCoy

    you are best to stick with the custom ovh kernel, its a custom kernel tweaked for the ovh dedicated hardware which the stock kernel wont have, you can try with the stock kernel from ubuntu’s site, but it would be best to stick with the one thats more geared toward the hardware ovh has

  7. Cacy

    I ‘m in the same kind of problem , but using CentOs . Does someone have done for CentOs ?
    What are the commands on CentOs to some the same problem, once there are no apt-get ?

    • nbeam

      Centos uses “yum” for its package manager. Debian and it’s variants (like Ubuntu) use “apt”. So commands in this article that talk about “apt” and/or reference a specific package name are not going to translate directly over to Centos unfortunately :(.

  8. ashishkarpe

    I am using Ubuntu on AWS ec2 instance so will it work for me ?

  9. Alberto Moratilla

    Awesome, you saved me a looooot of time 🙂

  10. Pingback: Differences between two Ubuntu Server 14.04.3 installations - Boot Panic

Join the discussion

Your email address will not be published. Required fields are marked *