ESXi hosts fail with purple diagnostic screen when connecting to the network I/O scheduler

Products

• VMware vSphere ESXi
• VMware vSphere ESXi 8.0

Issue/Introduction

In the backtrace, you see errors such as:
2024-08-05T02:57:57.302Z cpu20:2097484)0x45390a61b2a0:[0x420006423c31]RCU_WaitForGracePeriod@vmkernel#nover+0x42 stack: 0x1000000, 0x300000100, 0xbad0014, 0x420006518bc2, 0x0
2024-08-05T02:57:57.302Z cpu20:2097484)0x45390a61b2d0:[0x420006518bc1]vmk_HashRelease@vmkernel#nover+0x66 stack: 0x420006ac6eb5, 0x420007175487, 0x516574616572436b, 0x4307194064d0, 0x3030306
22024-08-05T02:57:57.302Z cpu20:2097484)0x45390a61b2f0:[0x420007175486]NetSchedHClkConnect@(netsched_hclk)#<None>+0x6ab stack: 0x30303062, 0x4300d3f27f48, 0x43126525a638, 0x0, 0x26

Environment

VMware ESXi 7.0

VMware ESXi 8.0

Cause

In very rare cases, the network I/O scheduler might hold a lock for a certain grace period in case of memory allocation failure during the connection. This lock might cause ESXi hosts to fail with a purple diagnostic screen and an error such as VERIFY bora/vmkernel/sched/cpushed.c.

Resolution

This issue is fixed in ESXi 7.0 Update 3v and VMware ESXi 8.0 Update 3e.

Release note for ESXi 7.0 U3v.

Release note for ESXi 8.0 U3e.