SOLVED Kernel conflicts on OVH servers due to missing exclude line in /etc/yum.repos.d/OVH-kernel.repo

[Rogerio]

Hello everyone,

OVH support is horrible, I had the same problem on four servers with CentOS 7. They replied (2 days later) that even being the OVH kernel, they do not support distributions and cannot help since "it's a unmanaged server"... OVH is very good, stable and good prices, but have the worst support ever. I use them only for non-critical services, because I cant trust them in emergency situations.

When purchasing a new dedicated server in OVH, choose to install the native kernel and not the OVH kernel.

Anyway, to stop the error, edit the file:

vi /etc/yum.repos.d/OVH-kernel.repo

and set:

enabled = 0

 

[JIKOmetrix]

Hello,

I'm testing this now with one of the servers.

- Mike

 

[JIKOmetrix]

Hello,

On host86 I ddi the Yum exclude, clean and update. That worked that time.

I'll try the other options too so see how that fares, I'm guessing they both accomplish the same task of excluding the kernal updates automatically.

We do manually kernel updates anyway.

Thanks for your help.

- Mike

 

[JIKOmetrix]

Hello,

I'd have to say with OVH you get what you pay for. They are very strict and have a no software support policy. You have to be your own system admin. That is what makes their prices so good. The infrastructure and pricing and features are hard to beat.

You just have know this going in to it when signing with OVH. Once you get your head around that and find others for support like cpanel forum and linux forums, it is not as bad as horrible. It is actually very good.

Also, I'd recommend the opposite regarding OVH kernels. If you install the native kernel you may in for a world of hurt looking for drivers and other items to get the system to boot.

Better to go with OVH kernels and manually update from time to time.

My 2 cents on OVH. If you use them for what they are good for then you'll have no complaints. They have always been receptive to replacing failing hard drives or raid controllers.

- Mike

Hello everyone,

OVH support is horrible, I had the same problem on four servers with CentOS 7. They replied (2 days later) that even being the OVH kernel, they do not support distributions and cannot help since "it's a unmanaged server"... OVH is very good, stable and good prices, but have the worst support ever. I use them only for non-critical services, because I cant trust them in emergency situations.

When purchasing a new dedicated server in OVH, choose to install the native kernel and not the OVH kernel.

 

[Rogerio]

Also, I'd recommend the opposite regarding OVH kernels. If you install the native kernel you may in for a world of hurt looking for drivers and other items to get the system to boot.

Better to go with OVH kernels and manually update from time to time.

Well, I don't agree. Better use a known and world-wide used kernel. As you can see, a lot of people are having problems, and this kind of problem (OVH kernel repo conflicts) is a way to destroy your server.

My 2 cents on OVH. If you use them for what they are good for then you'll have no complaints. They have always been receptive to replacing failing hard drives or raid controllers.

You're luck. Last time I have this kind of problem, they took 2 days to reply the ticket, and more 3 days to replace parts... Not even a good status system do they have ... just something like OVH Tasks (very poor).

Anyway, let's end this off topic.

 

[jvgn]

Hello.
I run yum exclude, clean and update and also worked. In my servers.

I think that OVH dont want you to use other Kernels because if your own kernel has problems (malware for example), it is also a problem for them. But they should verify these cases.

Thanks to all of you.

 

[pasanja]

I have the same problem with a server and another identical one that works on CENTOS 7.6

On the one that does not work, / boot/initramfs has been deleted and grub not updated during the yum update. At the next reboot, it displays the grub prompt because it tries to boot to a kernel that no longer exists.

After analyzing the server that works, it appears that the exclude line is missing in the OVH repo:

nano /etc/yum.repos.d/OVH-kernel.repo

[ovhkernel]
name=OVH kernel repo
enabled=1
baseurl=http://last-public-ovh-kernel.snap.mirrors.ovh.net/ovhkernel-rpm/
#gpgcheck=1
gpgcheck=0
gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-OVH
exclude=kernel-4.19-ovh-xxxx-std-ipv6-64-headers,kernel-4.19-ovh-xxxx-std-ipv6-64-hz1000-headers,kernel-4.19-ovh-mod-std-ipv6-64-headers

This restores kernel updates and cleanly reinstalls the latest version with its initramfs.

For those on the prompt grub, you have to restart in netboot and update grub
grub2-mkconfig -o /boot/grub2/grub.cfg
grub2-mkconfig -o /boot/efi/EFI/centos/grub.cfg

It seems to be an ovh error in its installation template.

 

[SS-Maddy]

Few more things worth mentioning in this thread as it was a result of a kernel update (client updated cPanel for patching exim exploit) and a OVH server.

Situation : Kernel was updated, rebooted the server and didn't come back. Client' server was EFI + Grub2 + no IPMI.
Initial check to do :

1. Check whether it is an EFI or normal BIOS
2. Check whether it runs grub2 or grub.
3. Check whether your server has IPMI enabled. Troubleshooting becomes simpler if it is.

Troubleshooting steps (need not be in order)

1. Try to boot from netboot. You are lucky, if it can run from netboot.
2. If that fails, boot to rescue
3. cat /etc/fstab and if it is UEFI firmware, check for a efi partition and mount it inside the /boot/efi. And inside check for grub.cfg. In this particular case grub.cfg was missing. If it's BIOS based check for /boot/grub2/grub.cfg
4. Also check for initramfs, if that is missing, you have to do some more extra steps so that you can build
5. Again check for /etc/default/grub to make sure that the OVH kernel is selected and not the "saved".

 

[dsys]

I have the same problem with a server and another identical one that works on CENTOS 7.6

On the one that does not work, / boot/initramfs has been deleted and grub not updated during the yum update. At the next reboot, it displays the grub prompt because it tries to boot to a kernel that no longer exists.

After analyzing the server that works, it appears that the exclude line is missing in the OVH repo:

nano /etc/yum.repos.d/OVH-kernel.repo

[ovhkernel]
name=OVH kernel repo
enabled=1
baseurl=http://last-public-ovh-kernel.snap.mirrors.ovh.net/ovhkernel-rpm/
#gpgcheck=1
gpgcheck=0
gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-OVH
exclude=kernel-4.19-ovh-xxxx-std-ipv6-64-headers,kernel-4.19-ovh-xxxx-std-ipv6-64-hz1000-headers,kernel-4.19-ovh-mod-std-ipv6-64-headers

This restores kernel updates and cleanly reinstalls the latest version with its initramfs.

For those on the prompt grub, you have to restart in netboot and update grub
grub2-mkconfig -o /boot/grub2/grub.cfg
grub2-mkconfig -o /boot/efi/EFI/centos/grub.cfg

It seems to be an ovh error in its installation template.

I had this conflict issue after WHM upgraded to v80.0.18 on CentOS 7.6 and just confirmed that the quoted solution above worked for one of my servers!

The issue:

[2019-06-22 00:03:21 +0000] [/usr/local/cpanel/scripts/rpmup] file /usr/include/asm/..install.cmd conflicts between attempted installs of kernel-4.19-ovh-mod-std-ipv6-64-headers-4.19_ovh_mod_std_ipv6_64-732421.x86_64 and kernel-4.19-ovh-xxxx-std-ipv6-64-headers-4.19_ovh_xxxx_std_ipv6_64-769503.x86_64

The solution that worked for me:
- Boot using OVH latest network kernel
- nano /etc/yum.repos.d/OVH-kernel.repo and add the exclude= line to the bottom
- yum update (updated kernel to latest)
- Run both grub2-mkconfig commands
- Change netboot back to HDD
- Reboot

Success!

Thank you @pasanja!

 

[WebCraker]

Hello everyone,

OVH support is horrible, I had the same problem on four servers with CentOS 7. They replied (2 days later) that even being the OVH kernel, they do not support distributions and cannot help since "it's a unmanaged server"... OVH is very good, stable and good prices, but have the worst support ever. I use them only for non-critical services, because I cant trust them in emergency situations.

When purchasing a new dedicated server in OVH, choose to install the native kernel and not the OVH kernel.

Anyway, to stop the error, edit the file:

vi /etc/yum.repos.d/OVH-kernel.repo

and set:

enabled = 0

Thanks alot, that worked for me. I edited the file and did a yum update and everything went fine.

But the question now is that going to make any issues when restarting the server as the server already using the network to boot (BootOVH kernel (includes up-to-date intel microcodes) ) ??