IPMI sensor mismatch / IPMI sensor data record unavailable on host

  • 19 July 2019
  • 2 replies

Badge +1
Greetings to all,

After an update of the AOS from 5.8.2 to 5.10.5, the following alerts began to be received:

IPMI sensor mismatch
IPMI sensor data unavailable on host

This is in a cluster with 3 Lenovo HX host, and all hosts have the same configuration when the cluster was created with the Foundations, I have already reviewed the access to the network to the BMC port and all the IPs respond and the users and passwords with correct.

This cluster uses the ESXi 6.5 hypervisor with vCenter 6.5, the information in the ESXi on the IPMI has already been added. The 3 CVM's respond to the ping between them and the BMC ports.

Is there any parameter within the CVM that should be reviewed or from the PRISM console?

2 replies

Userlevel 3
Badge +5
Hi @oscarascoli

Can you login on ESXi, run the following command and check the IP address reported by the IPMI (in your case Lenovo XCC):

/ipmitool lan print

After that, login to any CVM, run the following command and check if the IPMI address for that host reported by ipmitool command is the same of IP reported by the ncli command.

$ ncli host list
Userlevel 3
Badge +4

The first thing to do in these situations is run a full NCC health check. It seems entirely possible to me those alerts are from a scheduled check which ran during the upgrade.

If the relevant check passes now, no more work is needed. If there’s some persistent issue the NCC results will give a bit more detail.

like mentioned above, you can directly use ipmitool on AHV or ESXi nodes in a Nutanix cluster, or ipmiutil on a Hyper-V node. The article Common BMC and IPMI Utilities and Examples shows the specific syntax to use for each host OS. The lan print command mentioned above is a good test of the basic connectivity reflected in those alerts.