Hide Forgot
Description of problem: It's not possible to add a hypervisor with lesser cpu family in an existing hosted engine environment. As an example, if you are having an environment configured with "Sandy Bridge" and if later if we are adding a hypervisor which supports upto "Westmere" family, then the installation will fails with error "Invalid CPU type specified". However there is no option to change the current environment cpu type to Westmere as the answer.conf lies in the shared storage. We can workaround it by manually providing hosted-engine --deploy --config-append <answer.conf>. However from RHV 4.0 , cpu type also exist in the vm.conf and HE VM is tarted with qemu-kvm -cpu <cpu-family> which is configured in the OVF_STORE. So the HE VM will fail to start in new hypervisor with error "unsupported configuration: guest and host CPU are not compatible:" Editing the HE VM cpu type is locked in the RHEV-M portal. So it's not possible to change the cpu type which is configured during the installation. Version-Release number of selected component (if applicable): ovirt-hosted-engine-setup-2.0.2.2-2.el7ev.noarch rhevm-4.0.4.4-0.1.el7ev.noarch How reproducible: 100% Steps to Reproduce: 1. Install a hosted engine environment. 2. Add a new hosted engine host which does not support the existing cpu family in the HE environment. 3. Installation fails with error "Invalid CPU type specified". If we workaround it manually by "--config-append" , it's not possible to edit the HE cpu family from portal so that the vm.conf get synced with new information. Actual results: Cannot change HE cpu family to accommodate a hypervisor with lesser cpu family Expected results: Should be able to add new hypervisor by setting the lowest cpu family of the cluster. Additional info:
For now this is not allowed since the VM may crash when moving to a host with a lower cpu family.
*** Bug 1431211 has been marked as a duplicate of this bug. ***
Workaround: lower the CPU type in the cluster definition and change the VM core number for the engine VM to force an OVF_STORE update. Reboot the engine VM to start with the new CPU.
(In reply to Simone Tiraboschi from comment #5) > Workaround: lower the CPU type in the cluster definition and change the VM > core number for the engine VM to force an OVF_STORE update. > Reboot the engine VM to start with the new CPU. This works fine and I was able to add the HE host with lower cpu family. I will document this in a KCS. However the answer.conf in the conf_volume is still having the old cpu family . Not sure if it create any issue.
BZ<2>Jira Resync