Resolving suspected power problems

Power problems can be difficult to solve. For example, a short circuit can exist anywhere on any of the power distribution buses. Usually, a short circuit will cause the power subsystem to shut down because of an overcurrent condition.

Note

With GPUs greater than 250W (such as AMD MI-25) and CPUs greater than 165W and 165W low TCase SKUs (8180, 8168, 6154, 6146, and 6144), full performance cannot be guaranteed and CPU throttling might occur for ambient temperatures above 30°C.

Complete the following steps to diagnose and resolve a suspected power problem.

1. Check the event log and resolve any errors related to the power.
   Note

   Start with the event log of the application that is managing the server. For more information about event logs, see Event logs.
2. Check for short circuits, for example, if a loose screw is causing a short circuit on a circuit board.
3. Remove the adapters and disconnect the cables and power cords to all internal and external devices until the server is at the minimum configuration that is required for the server to start.
   Note

   See System configuration to determine the minimum configuration for your server.
4. Reconnect all ac power cords and turn on the server. If the server starts successfully, reseat the adapters and devices one at a time until the problem is isolated.

If the server does not start from the minimum configuration, replace the components in the minimum configuration one at a time until the problem is isolated.