Oh man, that's the worst kind of 3 AM wake-up call! When you're staring at that 'CPU FAULT' alarm and every minute costs thousands, here's my 5-step emergency protocol before you even think about calling the OEM:

1. First, don't panic and do no harm - Take a deep breath. Rushing can make things worse. Check if there's a failover system or load balancer that can redirect traffic while you work.

2. Gather intel and document everything - Check the server's management console (iLO, iDRAC, IPMI) for detailed error logs. Look for temperature readings, power supply status, and any other hardware indicators. Take screenshots or notes of everything you see.

3. Attempt a controlled reboot - If possible, gracefully shut down applications first, then do a full power cycle. Sometimes CPU faults clear after a reboot. If the fault LED stays amber after reboot, you've got a real hardware issue.

4. Check cooling and power basics - Verify the server has proper airflow and isn't overheating. Check power supply indicators and make sure all connections are secure. Overheating can trigger false CPU faults.

5. Isolate and verify the fault - If you have redundant hardware, try swapping components (if you're comfortable). Check if the fault follows a specific CPU socket or if it's system-wide. This gives you concrete info for the OEM support call.

Only after these steps would I call the OEM - armed with specific error codes, temperature logs, and what I've already tried. This way, you're not just reporting 'server broken' but giving them actionable data to speed up resolution!