How to rebuid a failed boot kernel of RHEL 3

[Gary L]

Hi

I have four HP BLADE servers, two dual core CPUs, 8GB RAM, run RHEL 3u4, boot kernel are 2.4.21-27.ELsmp and 2.4.21-27.EL.
Last year, those four servers were rebooted due to some reasons. The default boot kerenel *.ELsmp could booted up servers but NIC connection had a big problem, I only could access system via console or iLO remote console. I rebooted those four server through boot devie *.EL, it worked. But those system ran half system resource, half memory 4GB and just shows one CPU. As you know previous system could showed two physical CPUs and four logical CPUs.

My boss let me fix this kind of problem soon.

How to let boot device *.EL run full system resource 8GB RAM and 4 logical CPUs?
or
How to rebuild failed boot device 2.4.21-27.ELsmp?

Any answers will be very appreciate.

Thanks a lot.

-G

[Ivan Ferreira · Accepted solution]

You must boot with the working kernel, copy the support pack or NIC drivers, reboot with the SMP kernel and install the PSP/drivers with that kernel up.

[Gary L]

Hi Ivan

Thank you very much for your fast reply and good suggestions, would you please provide me more details,

Currently, I could boot from *.EL kernel, but this way only got half sysrem resource but others things are all correct. Boot device *.ELsmp could got full system resource but NIC didn't work. According as your suggestion, Do I need follow below steps:

1. boot system via boot device *.EL (worked one).

2. Copy some support pack and/or NIC drivers.
Could you please provide me more details, which file(s) need be copied for NIC issue in *.ELsmp?

3. Reboot system via *.ELsmp boot kernel.

4. Install the PSP/drivers.
For PSP installation, I could through HP SmartStar CP, my version is 8.00
# cd /mnt/cdrom/compaq/csp/linux
# ./install800.sh
I remember, it will generate two HP's boot kerenl instead of default Redhat's HP****.ELsmp, HP****.EL. Did you mention above way from smartstart Cd installing PSP?

How to install the "drivers".

Thanks a lot.

-G

[Ivan Ferreira]

As your "major network problems" descriptions is not very precise, I think that the problem may be the network drivers not being loaded to your SMP kernel.

>>> Could you please provide me more details, which file(s) need be copied for NIC issue in *.ELsmp?

That is the support pack files, of ir you have it on CD, continue with next step.

3. Reboot system via *.ELsmp boot kernel.

>>> 4. Install the PSP/drivers.

Yes, but as you probably have it installed, it should report which packages are already installed. The idea is to load the network drivers on your system, so force the installation of the network drivers.

The true is, I don't know if your "major network problem" is related to the missing driver/module on the kernel. I need the output of ifconfig command when the system is not working.

You should always have a good backup of the operating system before continue.

[Gary L]

Thanks Ivan.

I will provide your error details late.

Have a good day and thanks again for your help.

-G

[Nuwan Alwis]

Hi Gary,
Sorry if im kind of jumping out from no where. Any way just wanted to know how you decided that a kernel recompilation is required?

According to what i know
You can use the SMP kernel on the media(EL3 DVD) and copy it to /boot, configure the grub and you are ready to go.

A recompilation will required if your failed kernel had any drivers or other options specially configured to your server then u need to do so.
Check this if u need to add any drivers and recompile a kernel for your use

http://www.redhat.com/docs/manuals/linux/RHL-7.3-Manual/custom-guide/ch-custom-kernel.html

Hope this will be helpful and my friend Ivan can also comment on this

Good Luck!

[Gary L]

Hi Nuwan Alwis

Thank you very much for your good suggestion and explanations.

I will try your way soon.

Have a good day.

-Gary