One or more ESXi hosts may PSOD with stack trace containing the method NFS41_SMEventCreate
PSOD backtrace looks similar to below events:
PanicvPanicInt@vmkernel#nover+0x439
Panic_vPanic@vmkernel#nover+0x23
vmk_PanicWithModuleID@vmkernel#nover+0x41
NFS41_SMEventCreate@(nfs41client)#<None>+0x7d
NFS41_ConnectionRPCEventCB@(nfs41client)#<None>+0x1b
SunRPC_MsgCntxtRun@com.vmware.sunrpc#1+0x5db
SunRPCMsgWorldFunc@com.vmware.sunrpc#1+0x35
CpuSched_StartWorld@vmkernel#nover+0x77
During NFSv4.1 Server failover, a certain code path fails to release lock when the ESXi receives a different Server Major number from the Server. The host does not PSOD immediately after the server failover, the events keep building in the queue until there is no more free memory available to allocate for the event.
The issue is resolved in vSphere 7.0 U3f and later releases
No workaround available. NFS4.1 server failovers may be unplanned, which is outside of user control.