Update on this.
Managed to get past this problem by enabling UEFI booting (Yup my mistake since I see that is a requirement now having looked again at the install instructions) however hit a further problem.
Before I explain further, I previously used VIRL for my home lab , yes it had its install challenges but absolutely nothing compared to the experience I am going through with CML2. Home lab server is a DELL R610, 24 core 25G ram , RAID array, 4 GB NICS, near on perfect for home lab when self financed.
Initial image I attempted for bare metal install is the latest: 'cml2_p_controller-2.2.2-52.el8.52.x86_64-39.iso'
The Installation failed doe to lack of support for my storage controller. Reading forums upon forums (I have been on this four days now) Centos 8.x removed support for legacy controllers. One of which was my model which is LSI SAS1068E PCI-EXPRESS Fusion-MPT SAS [1000:0058] (rev 08)
As a direct consequence of this I am no longer able to run VIRL in my home LAB.
I found a generic fix for adding missing driver support at the time of CENTOS 8.x installation here:
https://elrepoproject.blogspot.com/2019/08/rhel-80-and-support-for-removed-adapters.html
Absolutely brilliant article and video but I found the link to 'supported devices' was broken so was unable to search for my required download.
After many hours of trying to demine the RPM download I should use for my SAS controller I managed to establish that the file I needed was this one:
dd-mptsas-3.04.20-4.el8_3.elrepo.iso
I formatted a USB, named it OEMDRV , and used mac OSX to burn the image onto USB.
Watching the boot process the driver is found, I was then able to select the storage for installation later in install cycle.
However, the problem was not over. Next error:
Problem: conflicting requests
-nothing provides kernel(dma_direct_map_sg) = 0x54b9950c needed by kmod- mptsas-3.04.20-4.e18_3.elrepo.x86_64
-nothing provides kernel(dma_direct_unmap_sg) = 0xf3324f6e needed by kmod- mptsas-3.04.20-4.e18_3.elrepo.x86_64
press ENTER to exit:
And that's it, no other options. The install process starts from the beginning again.
To overcome this obstacle I installed alternative storage by way of a USB connected HDD. The installation completes, but I find the starting of nodes slow probably due to restricted Read/write HDD speeds.
My next thought was to install ESXi on the server. Managed to find an eval copy of ESXi6.5.0 - Installed onto server fine.
When I attempted to deploy the ova file was met with an error:
A required disk image was missing
I see other members have experienced the same problem.
https://learningnetwork.cisco.com/s/question/0D53i00000VFp99/im-trying-to-install-cml-20-on-esxi-65when-i-deploy-the-ova-i-am-getting-a-required-disk-image-was-missingalso-the-cause-iscause-typeerror-scopewizardoptionsdataovfdatavirtualsystem-is-unde...
Besides these avenues, I have tried to install the following bare metal ISOs CML versions:
cml2_p_controller-2.2.1-36.el8.36.x86_64-36 - Simply fails early on in the process
cml2_p_controller-2.1.0-17.el8.x86_64-8 - almost installs, take an age, then fails to boot CML2
In summary , so far, I am only able to install CML2 version cml2_p_controller-2.2.2-52.el8.52.x86_64-39 using an external USB HDD as storage.
My question now is, am I alone in this install experience? and given CML2 (personal edition) is likely to be used in engineers home labs (often financed by themselves) why would legacy driver support be removed!?
I virtually have no way of using this CML software as it stands!
