Homelab Unexpected Power Outage Recovery Log: Complete Resolution Process for vCenter Startup Failure no healthy upstream

A reliable server will always teach you a lesson when you least expect it.

What happened

On an ordinary night, the power at home went out unexpectedly. I thought it was a minor issue—once the power came back, the server would naturally start up again. But when I went to check the next morning, I found it wasn't that simple.

• 半夜自动关机: The server briefly started after the power came back, but it automatically shut down again in the early hours of the morning.

• No response when pressing the power button: Tried to power on again; the CPU fan spun a few times and then stopped, and pressing the power button again produced no response.

• Fault scope: The motherboard indicator light and the network card indicator light are both flashing normally, indicating that the power supply standby circuit is good, and the problem lies in Signaling Link.

Phase 1: Hardware-level troubleshooting

Power supply protection mode triggered

phenomenon Press the power button; the CPU fan spins a few times and then stops, and the power supply fan does not keep running.

Troubleshooting process:

• Completely discharge: Unplug the power cord, press and hold the power button for 30 seconds to discharge residual power from the motherboard.

• Start minimized: Keep only the motherboard, CPU, one memory module, and the power supply; disconnect all hard drives and peripherals.

• Observe phenomena: The power supply fan spins briefly and then stops, indicating that the motherboard has entered protection mode.

CMOS reset

phenomenon After minimal boot failure, the problem may be in the BIOS settings.

Operation:

• Unplug the power cord and locate the coin cell battery (CR2032) on the motherboard.

• Remove the battery and wait 2 minutes.

• Reinstall the battery, plug it back in, and power it on.

result: The server started automatically, but the system disk could not be recognized (ESXi did not boot).

Phase 2: System startup failed

Boot disk not found

phenomenon: The server can be powered on, and the fans are spinning normally, but it cannot boot into the ESXi system. Ping 192.168.0.201 Timeout.

Cause Analysis: After the BIOS reset, the boot order may have changed, or the SATA cable may have come loose during the operation.

Solution:

• Reconnected the SATA data cable and power cable of the system drive (Samsung 860 EVO).

• Restart the system by shutting down and rebooting to allow the BIOS to automatically detect the system disk.

Result: ESXi started successfully,192.168.0.201 It can be accessed normally.

Phase 3: vCenter startup failed

no healthy upstream

phenomenon:

• VAMI management port (:5480) Can be accessed normally, showing all core components with status "Normal".

• Main Web port (:443) prompt when accessing no healthy upstream.

Cause analysis: vCenter's core services have started, but the service responsible for providing the Web interface (httpd) did not respond normally.

Troubleshooting and Repair:

• Confirm service status via VAMI:

• Visit https://192.168.0.202:5480

• Confirm that components such as CPU, memory, database, and storage are all in green normal status.

• Go to the vCenter console and manually start the service.:

• Log in to the vCenter virtual machine console (using root account).

• Run the command to check the service status:

service-control --status

• Discovery vpxd,postgres and other key services failed to start properly.

• Manually start all vCenter services:

service-control --start --all

• Wait about 2-3 minutes until all services have started successfully.

• Verify fix:

• Revisit https://192.168.0.202, vCenter Web client displays normally.

• Use [email protected] Successfully logged in.

Full repair timeline

time point	phenomenon	Operation	Result
凌晨（断电后）	Server automatically shuts down	-	Fault occurrence
Morning	No response when pressing the power button	Fully discharge, minimize startup	The fan stops after spinning a few turns.
Morning	Still unable to boot normally.	CMOS reset	The server is powered on, but the system disk cannot be recognized.
Noon	ESXi cannot boot	Reconnect the SATA cable and manually select the boot drive.	ESXi successfully started
afternoon	vCenter Web interface no healthy upstream	Execute via the console service-control --start --all	vCenter Complete Recovery

Experience Summary and Recommendations

1. The impact of power outages on Homelab cannot be underestimated.

A seemingly ordinary power outage may trigger a series of chain reactions:

• The power protection mechanism was triggered, preventing the device from powering on.

• BIOS settings lost, boot order changed.

• vCenter services have complex dependencies and may not recover automatically after an abnormal shutdown.

2. Standard approach to solving this type of problem

Stage	Problem manifestation	Priority troubleshooting direction
Hardware layer	No response when pressing the power button; the fan spins briefly and then stops.	Power protection mode, CMOS reset, minimal boot
System layer	The server can be powered on, but ESXi cannot boot.	BIOS boot order, SATA cable connection status
Service Layer	ESXi starts normally, but vCenter is inaccessible.	VAMI（:5480）Check service status and manually start the service

3. A few suggestions for reducing future risks

• Configure VMware Tools to start automatically: In the vCenter virtual machine options, set "Automatic Startup" to "Always start automatically".

• Consider UPS (uninterruptible power supply) If conditions permit, a UPS can provide servers with buffer time after a power outage, avoiding frequent abnormal shutdowns.

• Regularly record system status: Record the IP addresses, passwords, and service status of ESXi and vCenter for quick identification during troubleshooting.

Conclusion

Although the process of this power outage repair was full of twists and turns, it also gave me a deeper understanding of the Homelab's hardware and service architecture. Often, when things go smoothly, we don't notice the fragility of the system, and a failure is often the best learning opportunity.

This server is now running quietly again, continuing to support my K8s experiments, GitLab CI/CD, and distributed R&D environment. And I also hope this record can, someday in the future, help you get out of a similar predicament more quickly.

Related documents

• First in the series:Homelab: From Zero to One Hands-on Series- This series topic entry

• Fundamentals:Building a Homelab from Scratch (Part 1): Creating an ESXi 8.0 Bootable USB Drive in macOS- Create ESXi installation media from scratch

• Advanced Remote Access:The Most Elegant Homelab Remote Access Solution: Cloudflare Tunnel Zero Trust Practice- No public IP required; securely access internal network services

• Ultimate Architecture:Homelab Ultimate Remote Access Architecture: Cloudflare Tunnel + Nginx Unified Access Layer- Unified access layer, solving the challenges of subdomain sprawl and SSH traversal

Appendix: IP Addresses Involved

• ESXi management address:https://192.168.0.201

• vCenter management address:https://192.168.0.202

• vCenter VAMI management port:https://192.168.0.202:5480