4.X kernel on RHEL7 x86_64

Latest response

Does anyone know how to get/install the kernel-alt package on a x86_64 system? I know that it is not officially supported, but this is for a test/development environment that requires the use of the 4.X kernel.

I have it working with CentOS7 (1804) and the kernel-ml from elrepo.org, and now would like to see if I can replicate a similar setup with RHEL7. Long-term this project will be built using RHEL (after 4.X is officially supported), so I would prefer to begin with RHEL instead of CentOS if possible.

Or is it possible to side-load / source compile a new kernel on RHEL?

Thanks in advance

Responses

ELRepo's kernel-ml and kernel-lt are built for RHEL. kernel-ml is fast moving but kernel-lt should remain at 4.4 for a while. Will this not work for you?

kernel-lt would work just fine, but I thought that RHEL blocked access to 3rd-party repos to ensure stability and regulatory compliance. Is there any documentation on how to load an unofficial kernel, either through yum or manually side-loading? Just trying to save myself some time instead of re-inventing the wheel.

I figured it out. Now to rebuild the installation disc with a kickstart file and the new kernel...

Glad that you figured it out. For posterity, I briefly summarize the steps for those who may be interested:

(1) Set up the elrepo repository by following the instructions at ELRepo homepage. (2) To install kernel-lt, run

yum --enablerepo=elrepo-kernel install kernel-lt 

We have some cutting edge hardware (latest AMD chipset) and are having issues with NVMe timeouts on the 3.10 kernel. We require FIPS to function, which does on the 3.10 kernel. When we try the 4.4 kernel we get transform failures for tcrypt and stops running with a FIPS integrity test failure. We have done a dracut -f when running the 4.4 kernel without FIPS. When the system is up on the 4.4 kernel without FIPS the NVMe errors have stopped. Any ideas?

Are you seeing nvme errors such as:

Jan 25 10:43:46 myhostname kernel: [ 9278.208176] nvme nvme1: controller is down; will reset: CSTS=0xffffffff, PCI_STATUS=0xffff
Jan 25 10:43:46 myhostname kernel: [ 9278.208180] sched: RT throttling activated
Jan 25 10:43:46 myhostname kernel: [ 9278.208176] nvme nvme1: controller is down; will reset: CSTS=0xffffffff, PCI_STATUS=0xffff
Jan 25 10:43:46 myhostname kernel: [ 9278.208180] sched: RT throttling activated
Jan 25 10:43:46 myhostname kernel: nvme nvme1: controller is down; will reset: CSTS=0xffffffff, PCI_STATUS=0xffff
Jan 25 10:43:46 myhostname kernel: nvme nvme1: controller is down; will reset: CSTS=0xffffffff, PCI_STATUS=0xffff
Jan 25 10:43:46 myhostname kernel: sched: RT throttling activated
Jan 25 10:43:46 myhostname kernel: sched: RT throttling activated

And then the system locks up?

Did the 4.4 or 4.? solve these problems?

Thanks.

Alexander,

I am not an expert in NVMe, but I understand that many fixes have gone into the Red Hat kernel for NVMe. Does the latest Red Hat 3.10 kernel solve the NVMe timeouts? If you still have problems with NVMe, and you can open a support case, report the issue.

I have the same problem and it seems the problem might be because CONFIG_CRYPTO_FIPS=y seems to be missing from elrepo's kernels

You can file a request at ELRepo's bug tracker. Usually a config change like that will be made in the next update release of the kernel.

Hi Team ,

Is this posible to upgrade the kernel to 4 in redhat linux .

Hi Victor,

Kernel 4 possible ? Yes ! (by installing the ELRepo kernel) ... officially supported by Red Hat ? No !
But once the final edition of RHEL 8 gets released, you'll have kernel 4 included out-of-the-box. :)

Regards,
Christian