After running the Linux fstrim command on a virtual machine, the ESXi host fails with a purple diagnostics screen

Products

• VMware vSphere ESXi 8.0

Issue/Introduction

ESXi host fails with a purple diagnostic screen (PSOD).

From the vmkernel.logs we can see error similar to:

vmkernel.log:YYYY-MM-DDTHH:MM:SS.XXXZ In(182) vmkernel: cpu0:1000210432)PB3: ClearNLRUCacheSlowPath:3202: PB3 Clear NLRU slow path, numPBs 1 unlinked PBs 1
vmkernel.log:YYYY-MM-DDTHH:MM:SS.XXXZ In(182) vmkernel: cpu0:1000210432)PB3: ClearNLRUCache:3252: [STRESS] PB3ClearNLRUCache: Fake tempPBCacheElems allocation failure
vmkernel.log:YYYY-MM-DDTHH:MM:SS.XXXZ Wa(180) vmkwarning: cpu0:1000210432)WARNING: PB3: ClearNLRUCache:3262: No memory for clearing NLRU Cache 0x431cd1e888c0 with 1 numPBs
vmkernel.log:YYYY-MM-DDTHH:MM:SS.XXXZ In(182) vmkernel: cpu0:1000210432)PB3: ClearNLRUCache:3354: Clear NLRU cache encounter error Out of memory, num PBs 1 attempting slow path
vmkernel.log:YYYY-MM-DDTHH:MM:SS.XXXZ In(182) vmkernel: cpu0:1000210432)PB3: ClearNLRUCacheSlowPath:3202: PB3 Clear NLRU slow path, numPBs 1 unlinked PBs 1
vmkernel.log:YYYY-MM-DDTHH:MM:SS.XXXZ In(182) vmkernel: cpu1:1000210432)PB3: ClearNLRUCache:3252: [STRESS] PB3ClearNLRUCache: Fake tempPBCacheElems allocation failure
vmkernel.log:YYYY-MM-DDTHH:MM:SS.XXXZ Wa(180) vmkwarning: cpu1:1000210432)WARNING: PB3: ClearNLRUCache:3262: No memory for clearing NLRU Cache 0x431cd205af30 with 3 numPBs
vmkernel.log:YYYY-MM-DDTHH:MM:SS.XXXZ In(182) vmkernel: cpu1:1000210432)PB3: ClearNLRUCache:3354: Clear NLRU cache encounter error Out of memory, num PBs 3 attempting slow path
vmkernel.log:YYYY-MM-DDTHH:MM:SS.XXXZ In(182) vmkernel: cpu1:1000210432)PB3: ClearNLRUCacheSlowPath:3202: PB3 Clear NLRU slow path, numPBs 3 unlinked PBs 3
vmkernel.log:YYYY-MM-DDTHH:MM:SS.XXXZ In(182) vmkernel: cpu1:1000210432)PB3: ClearNLRUCache:3252: [STRESS] PB3ClearNLRUCache: Fake tempPBCacheElems allocation failure

Environment

ESXi 8.0 Update 3f or earlier.

Cause

A VMFS heap memory exhaustion might cause memory access violations when clearing the pointer block cache.

As a result, when you run the Linux fstrim command on a virtual machine, the ESXi host might fail with a purple diagnostics screen with a PB3Manager error in the backtrace.

Resolution

This issue is resolved in this release. ESXi 8.0 Update 3g

For more information, refer to VMware ESXi 8.0 Update 3g Release Notes

Additional Information

For vSphere Quick Boot support in ESXi 8.0 Update 3g, see the list for each vendor in the Additional Information section of the Understanding ESXi Quick Boot Compatibility KB.