PowerEdge: "No Memory Found" event is displayed on NVIDIA Bluefield 3 enabled servers during POST
Summary: A recent combination of Data Processing Unit (DPU) firmware, Complex Programmable Logic Device (CPLD) firmware, and iDRAC enabled a graceful shutdown feature. Performing certain tasks can cause a "No Memory Found" event during boot. ...
Symptoms
A recent combination of Data Processing Unit (DPU) firmware, Complex Programmable Logic Device (CPLD) firmware, and iDRAC enabled a graceful shutdown feature. Performing certain tasks can cause a "No Memory Found" event during boot. Examples of these tasks such are BIOS updates, PCIe Switch Board (PSB) updates, BIOS configuration changes, or server cold reboot. Then on the subsequent reboot the server may encounter this error.
This feature is enablement of graceful shutdown of the DPU.
This impacts all DPU enabled platforms.
- R660
- R760
- R760XA
- XE9680
This impacts NVIDIA BlueField-3 DPUs with DPU mode enabled.
- NVIDIA Bluefield-3 B3210e
- NVIDIA Bluefield-3 B3220
Note: The NVIDIA Bluefield-3 B3140H comes set in Super NIC mode. If it is changed to DPU Mode, it can also encounter this event.
Example of error during Power On Self-Test (POST):
Example of error in the Lifecycle Log:
Cause
When this new feature was added, the functionality required changes to the NVIDIA BlueField-3, the iDRAC, and the CPLD firmware of the affected platforms. During the power cycle process after certain tasks, the timing of the various state machines are not synchronized, leading to the event.
Note: This is done when the iDRAC has found at least one DPU running in DPU mode. If a DPU is running in Super NIC mode, then it does not encounter this.
Impacted CPLD versions:
- 1.1.5
- 1.1.7
- 7.10.50.00
- 32.40.1000
Resolution
When No Memory Detected Happens:
The server remains at the error message status in the BIOS, as shown in the screenshot above. The server sits in this state and waits for approximately 1 minute, then it reboots automatically. Once the server reboots, it is fully functional and memory is seen as expected.