I have a fully patched (I've tried with various levels of patches) ESXi 5.5 server running the following -
-> Ubuntu 14.04 via a Broadcom NIC
-> Windows 2012R2 via a Intel 82541PI Gigabit ethernet controller - host OS is using VMXNET3 driver
-> Windows 2012R2 via another Intel 82541PI Gigabit ethernet controller - host OS is using VMXNET3 driver
Two or three times every 24 hours I get the PSOD -
2015-06-15T12:38:41.246Z cpu3:35592)VMware ESXi 5.5.0 [Releasebuild-2718055 x86_64]
PCPU 4: no heartbeat (2/2 IPIs received)
2015-06-15T12:38:41.246Z cpu3:35592)cr0=0x80050031 cr2=0xba631eeff8 cr3=0x1910f1000 cr4=0x42668
2015-06-15T12:38:41.246Z cpu3:35592)pcpu:0 world:32787 name:"uplinkLoadBalancerWorld" (S)
2015-06-15T12:38:41.246Z cpu3:35592)pcpu:1 world:33466 name:"tq:rdt" (S)
2015-06-15T12:38:41.246Z cpu3:35592)pcpu:2 world:32860 name:"CpuSchedPeriodic" (S)
2015-06-15T12:38:41.246Z cpu3:35592)pcpu:3 world:35592 name:"vmm1:*****-WIN2012R2" (V)
2015-06-15T12:38:41.246Z cpu3:35592)pcpu:4 world:33441 name:"net-lacp" (U)
2015-06-15T12:38:41.246Z cpu3:35592)pcpu:5 world:32962 name:"memMap-5" (S)
2015-06-15T12:38:41.246Z cpu3:35592)pcpu:6 world:35590 name:"vmm0:*********-WIN2012R2" (V)
2015-06-15T12:38:41.246Z cpu3:35592)pcpu:7 world:32800 name:"PathTaskmgmtWatchdog" (S)
2015-06-15T12:38:41.246Z cpu3:35592)@BlueScreen: PCPU 4: no heartbeat (2/2 IPIs received)
2015-06-15T12:38:41.246Z cpu3:35592)Code start: 0x418015000000 VMK uptime: 0:04:17:08.008
2015-06-15T12:38:41.246Z cpu3:35592)Saved backtrace from: pcpu 4 Heartbeat NMI
2015-06-15T12:38:41.246Z cpu3:35592)0x41238a85d450:[0x4180157cc93a]e1000_intr@<None>#<None>+0x2e stack: 0x412300001017
2015-06-15T12:38:41.247Z cpu3:35592)0x41238a85d490:[0x41801568fd7e]Linux_IRQHandler@com.vmware.driverAPI#9.2+0x2a stack: 0x4108400c5660
2015-06-15T12:38:41.247Z cpu3:35592)0x41238a85d520:[0x41801506b0b6]IRQ_DoInterrupt@vmkernel#nover+0x33e stack: 0x4
2015-06-15T12:38:41.247Z cpu3:35592)0x41238a85d560:[0x418015064303]IDT_IntrHandler@vmkernel#nover+0x12b stack: 0x41238a85d640
The host will change between each 2012R2 server, it is never the Ubuntu server (also using VMXNET3 driver).
As far as I can find the solution is either use the E1000 will no RXpolling or use the VMXNET3 - I've tried both with the same results.
The NIC under vswitch properties is still using E1000 - should / can this changed?
Any help / feedback great appreciated!
As per VMWare's HCL the Intel 82541PI is only compatible up to VMWare 3.5: VMware Compatibility Guide: I/O Device Search
So it's quite possible that the latest e1000 driver for the physical NIC is causing issues.
What brand and model is that server? Is the NIC probably vendor supported (e.g. by HP)?
It's a home built server for a Lab etc. I'll order a different card(s) and check the HCL a little more thoroughly this time.
Thanks for the reply!