ESXi/ESX host using Emulex 10Gb NIC cards fails with a purple diagnostic screen with the error: PCPU #: no heartbeat
Article ID: 310135
Updated On:
Products
VMware vSphere ESXi
Issue/Introduction
Symptoms:
- An ESXi/ESX host becomes unresponsive and displays a purple diagnostic screen indicating that a CPU did not receive a heartbeat.
- The purple diagnostic screen is triggered by a non-maskable interrupt (NMI).
- You see diagnostic information similar to:
cpu7:xxxxxxx)@BlueScreen: PCPU 10: no heartbeat (2/2 IPIs received) cpu7:xxxxxxx)Code start: xxxxxxxxxxxxxx VMK uptime: 26:05:22:00.554
cpu7:xxxxxxx)Saved backtrace from: pcpu 10 Heartbeat NMI
cpu7:xxxxxxx)0x4122457dba78:[xxxxxxxxxxxxxx]Util_Udelay@vmkernel#nover+0x2d stack: 0x418000010000
cpu7:xxxxxxx)0x4122457dbac8:[0x4180260e3e56_be_mpu_post_wrb_ring@<None>#<None>+0xed stack: 0x41001c401150
cpu7:xxxxxxx)0x4122457dbb28:[0x4180260e0649 be_function_post_mcc_wrb@<None>#<None>+0x128 stack: 0x0
cpu7:xxxxxxx)0x4122457dbb78:[xxxxxxxxxxxxxx]be_get_die_temperature@<None>#<None>+0xb7 stack: 0x41224b0e7000
cpu7:xxxxxxx)0x4122457dbcc8:[xxxxxxxxxxxxxx]rate_timer_func@<None>#<None>+0x66b stack: 0x4122457dbd18
cpu7:xxxxxxx)0x4122457dbd68:[xxxxxxxxxxxxxx]Timer_BHHandler@vmkernel#nover+0x226 stack: 0xfffc013000ffff
cpu7:xxxxxxx)0x4122457dbde8:[xxxxxxxxxxxxxx]BH_Check@vmkernel#nover+0x98 stack: 0x2457dbe28
cpu7:xxxxxxx)0x4122457dbed8:[xxxxxxxxxxxxxx]CpuSchedDispatch@vmkernel#nover+0x11ea stack: 0x0
cpu7:xxxxxxx)0x4122457dbf48:[xxxxxxxxxxxxxx]CpuSchedWait@vmkernel#nover+0x242 stack: 0x410000000000
cpu7:xxxxxxx)0x4122457dbf98:[xxxxxxxxxxxxxx]CpuSched_VcpuHalt@vmkernel#nover+0x14b stack: 0x418025bb8619
cpu7:xxxxxxx)0x4122457dbfe8:[xxxxxxxxxxxxxx]VMMVMKCall_Call@vmkernel#nover+0x1af stack: 0x0 - The ESXi/ESX host is using Emulex 10Gb NIC cards.
Environment
VMware vSphere ESXi 5.1
VMware vSphere ESXi 5.0
VMware vSphere ESXi 5.5
VMware ESX 4.0.x
VMware ESX 4.1.x
VMware vSphere ESXi 5.0
VMware vSphere ESXi 5.5
VMware ESX 4.0.x
VMware ESX 4.1.x
Resolution
This is a known issue affecting Emulex 10 Gb Nics with older firmware versions.
To resolve this issue, upgrade 10 Gb NIC firmware to version 4.6.166.6105 or later and must use BE2NET drivers of version 4.6.166.9 or later. Both are available from the hardware vendor.
To work around this issue, perform these preemptive actions:
- Ensure that Spanning Tree Protocol is enabled on upstream switches. For more information, see STP may cause temporary loss of network connectivity when a failover or failback event occurs (1003804).
- For IBM servers, enable Management Network Auto-Discovery (MCAD) in the BladeCenter AMM.
- Ensure your physical switch firmware is up-to-date. For IBM Blade servers, ensure the firmware of the Virtual Fabric Switch is up-to-date.
Additional Information
Feedback
Yes
No
Powered by
