In last week I upgraded my laptop from T420s to E485 to get maximum 32G memory also because the T420s was so old and the cooling system worked inefficiently, so during the summer it would become slower and not stable.
I kept all the VirutualBox VMs as I did not want to install and configure them one more time and I thought the VMs would work without any issue as they were VIRTUAL machines, right?
I started one Oracle11g RAC node, checked the cluster services and confirmed it worked well ... just for several minutes. Suddenly I found the cursor was frozen not just for the VM but also for the host machine. I waited for some time and had to powered off the laptop by pressing the power button.
I did more tests and found not all of them got this trouble. One VM was installed as base template and it worked really well, but one fully cloned Oracle RAC VM on it had this problem.
I believed the Linux system should be compatible with the T420s and the E485 as the system itself was only some RPM packages so there must be something related with the hardware.
I noticed the Intel VT-x(T420s) and AMD AMD-V(E485) were not compatible, so I guessed during the Oracle installation, especially the compiling period, some binary files would be compiled making use of the CPU hardware virtualization features to get higher performance.
On the failed RAC node, I disabled the clusterware and rebooted it to hope it would work.
I run the 'ps -elf' frequently and found although the clusterware was stopped but the Oracle TFA was still running and soon it failed again.
Oracle TFA was used to collect and analyze the system performance information so maybe it was the root cuase, right? I disabled it also and rebooted the node one more time.
Checking and checking, yeah, the failed node did work well this time.
How to fix it? Relink the GRID_HOME and then I could start the clusteware and TFA as normal.