The New Server Won't Boot After Racking
A technician racks a new 2U server in a colo cage. The server has dual redundant 1600W power supplies. The rack PDU is a 208V three-phase unit with C13 outlets. The tech plugs both PSUs into the same PDU strip using two C13-to-C14 cords. On power-up, the server's front panel shows a steady amber light and no POST. The tech reseats the DIMMs and CPU heatsink, then tries again - same result. The server's iDRAC (management controller) is reachable and reports "Power Supply 2: Input Lost." The PDU strip's breaker has not tripped. The tech wants to open a vendor RMA for a dead motherboard.
1. Interpret the iDRAC message and identify the most likely root cause before any RMA.
Read the worked answer
iDRAC message: "Power Supply 2: Input Lost" means PSU 2 is not receiving AC input, even though PSU 1 is. The server is running on one PSU, which may be insufficient for full POST power draw - hence the amber light and no POST.
Most likely root causes (in order):
- Bad or unseated C13 cord on PSU 2, or the cord is plugged into a dead outlet on the PDU strip.
- PDU outlet failure - a single C13 outlet can fail even if the strip breaker is fine.
- PSU 2 itself is faulty - swap cords first to isolate.
- PSU 2 not fully seated in its bay.
Why not the motherboard: The iDRAC is alive and reporting, which means the management plane and standby power are working. A dead motherboard would typically not produce a clean "input lost" message. The amber light is consistent with a power fault, not a board fault.
Do not open an RMA yet - isolate the power path first.
2. Walk through the isolation steps to confirm the fault, and state what you'd do if it is the PSU.
Read the worked answer
Isolation steps:
- Swap cords between PSU 1 and PSU 2. If the fault follows the cord, replace the cord.
- Move PSU 2's cord to a known-good outlet on the PDU strip. If the fault clears, the original outlet is dead - tag it out.
- Swap PSU 1 and PSU 2 physically (if the model allows). If the fault follows the PSU, the PSU is bad.
- Check PSU seating - reseat PSU 2 firmly until it clicks.
- Check PDU strip load - verify the strip is not at capacity; a 1600W PSU pair on 208V draws significant current. Confirm the strip's rating and that both PSUs are not on the same branch circuit if redundancy is required.
If it is the PSU:
- Replace PSU 2 with a matching spare. Do not RMA the whole server.
- If no spare, run temporarily on PSU 1 only if the server's power draw is within one PSU's capacity, and schedule replacement.
- Redundancy note: both PSUs on the same PDU strip means a strip failure takes down the server. For true redundancy, PSU 1 and PSU 2 should be on separate PDUs / separate branch circuits (A/B feeds). Flag this to the colo team.
3. After the server boots, what cabling/configuration change would you recommend, and why?
Read the worked answer
Recommendation: Move PSU 1 to PDU A and PSU 2 to PDU B, on separate branch circuits, so a single PDU or circuit failure does not take down the server. This is the standard A/B feed redundancy pattern in colos.
Additional checks:
- Verify the PDU strips are on different upstream circuits (not just different outlets on the same strip).
- Confirm the server's power draw is within one PSU's capacity so it can survive a single-feed loss.
- Label both cords at both ends (source and destination) for future troubleshooting.
- Update the rack elevation and cable schedule documentation.
- If the colo only provides one feed to the cage, document the single point of failure and raise it with the colo provider.