ESX/ESXi host console reports the message: "Significant imbalance between NUMA nodes detected"

Products

• VMware vSphere ESXi

Issue/Introduction

Symptoms:

• When accessing the console of the ESX/ESXi host, you see a message similar to:
  
  TSC: 631030280 cpu0:0)NUMA:827 : Significant imbalance between NUMA nodes detected. Performance may be impacted.
  
  
• In the /var/log/boot.gz file, you see entries similar to:
  
  TSC: 290948260 cpu0:0)WARNING: NUMA: 1418: A significant memory (DRAM) imbalance (more than 30 percent) was detected between NUMA nodes 0(524232 MB) and 1(0 MB). This may impact performance
  WARNING: NumaCodeRep: 613:* Could not replicate kernel code on node 1*0:00:00:04.788 cpu0:1)
  WARNING: NumaCodeRep: 613: Could not replicate kernel code on node 20:00:00:04.788 cpu0:1)
  WARNING: NumaCodeRep: 613: Could not replicate kernel code on node 30:00:00:07.924 cpu0:32768)

Cause

This issue may occur after upgrading your ESX/ESXi host.

This message is reported when memory is not evenly distributed between processor sockets. The system is completely functional in this configuration but may run some workloads at reduced efficiency due to increased memory latencies.

Each CPU socket has its own RAM slots.

Note: Damaged or faulty physical RAM modules can also cause this issue.

Resolution

To resolve this issue, evenly balance RAM modules between processor sockets. That is, ensure that each bank of RAM slots has the same amount of RAM.

For example, if CPU socket 1 has a total of 8 GB of RAM, ensure that CPU socket 2 also has 8 GB of RAM.

Note : The cause could be a benign side effect from the vendor's implementation of our Reliable Memory feature. Ensure that memory on the nodes are balanced. For more information and assistance, contact your vendor.

For more information, see About Reliable Memory.