Disable ENS on ESXi host which is causing the host to PSOD

Products

• VMware vSphere ESXi

Issue/Introduction

PSOD occurring on ESXi hosts where ENS is enabled -

Output of the PSOD:

VMware ESXi 9.0.1 [Releasebuild-24957456 x86_64]
#PF Exception 14 in world 2105476:ensLBMain IP 0x4200312c03d2 addr 0x4
PTEs:0x8a2a6a0027:0x814b2aa027:0x0;

Module(s) involved in panic: [nsxt-ens-24952112 Version 2.0.0-0 RELEASEbuild-24952112]
cr0=0x8001803b cr2=8x4 cr3=0x6f0e4000 cr4=0x14216c
FMS=06/cf/2 uCode=0x210002b3
frame=0x453a69f9bb40 ip=0x4200312c03d2 err=0x0 rflags=0x10246
rax=0x3 rbx=0x45020f01e7d0 rcx=0x3
rdx=0x0 rbp=0x45020eb040a0 rsi=0x0
rdi=0x431a00e0e358 r8=0x0 r9=0x453a69f9ba3f
r10=0х0 г11=0х0 г12=0x0
r13=0x45020016b6f0 r14=0x45020d6f4490 r15=0x0
*PCPU49:2105476/ensLBMain
PCPU 0: SVVVVSVSVVUSVVSVVSVVSVSSVSSSVUSVVVVSVUVVUUVUSSVVUSSVUSVVSVVVVVVV
PCPU 64: SSUVVVVVVVSVSVSVSVUVVVVSVVVSVVVUSVVVVVVVVVVVVSVVVVVVVUVVVVVSVVVS
Code start: 0x42002ee00000 VMK uptime: 37:12:37:50.765
0x453a69f9bc00:[0x4200312c03d2]EnsNetQGenericPlanQueueOps@(nsxt-ens-24952112)#<None>+8x11b2 stack: 0x2
0x453a69f9bd30: [0x4200312c654b]EnsMetQProcessQueueOperations@(nsxt-ens-24952112)#<None>+0x304 stack: 0x45020eb040a0
0x453a69f9bdd0: [0x4200312c8784]EnsMetQProcessQueueOperationsA11@(nsxt-ens-24952112)#<None>+0x41 stack: 0x0
0x453a69f9be00: [0x4200312e5697]EnsLBProcPendingOperations@(nsxt-ens-24952112)#<None>+0x40 stack: 0x0
0x453a69f9bed0: [0x4200312e9960]EnsLBMain@(nsxt-ens-24952112)#<None>+0xlel stack: 0x45020016b560
0x453a69f9bfa0:[0x42882efbc859]vmkWorldFunc@vnkernel#nover+8x36 stack: 0x42002efbc855
0x453a69f9bfe0:[0x42002f514f77]CpuSched_StartHorld@vnkernel#nover+8xc4 stack: 0x0
0x453a69f9c000:[0x42002ef5824b]Debug_IsInitialized@vnkernel#nover+0x10 stack: 0x0
base fs=0x0 gs=0x42004c400000 Kgs=0x8
2026-02-14T04:30:57.862Z cpu28:2098140)vmsyslog logger 10.82.11.65:514 lost 22469 log messages
2026-02-14T04:30:56.796Z cpu28:2098140)vmsyslog logger vm-vcf-log.hargreaveslansdown.co.uk:514 lost 4327 log messages
2026-02-13T16:30:56.944Z cpu95:2098140)vmsyslog logger 10.82.11.65:514 lost 1 log messages
2026-02-13T16:30:56.628Z cpu107:2098140)vmsyslog logger vm-vcf-log.hargreaveslansdown.co.uk:514 lost 3 log messages
2026-02-11T15:15:31.445Z cpu86:11622320)BC: 3361: File host-21-hb closed with dirty buffers. Possible data loss.
026-02-11T15:15:29.455Z cpu64:11622202)BC: 3361: File host-21-hb closed with dirty buffers. Possible data loss.
2026-02-04T21:54:46.153Z cpu0:2098140)vmsyslog logger 10.82.11.65:514 lost 3912 log messages
2026-02-04T09:54:45.719Z cpu54:2098140)vmsyslog logger 10.82.11.65:514 lost 1 log messages
0:00:00:01.746 cpu0:2097152)LLSwap: 3832: Host Cache will be deprecated in the next najor release.
No disk partition configured to dump data.
Finalized dump header (18/18) FileDump: Successful.
No port for remote debugger.

Environment

VCF 9.x

Resolution

Steps to be carried out on ESXi host -

1. localcli system module parameters set -m bnxtnet -p "num_rss_pools=1,1,1,1" ->This assumes that the host has 4 bnxtnet uplinks, change to suit the amount of uplinks per host.
2. Reboot the host.

With this config option change, ENS only creates the defaultQ RSS and the host won't get into the code path that lead to the PSOD.

The impact should be minimal in terms of both functionality and performance as defaultQ has 8 RSS queues which should be enough even to drive NSX-T Edge VMs.