IPMI Discrete Sensors

[jrosetto]

I have been trying to get discrete sensors working with zabbix to no avail. The Zabbix documentation says I should be able to do it.

I would like to monitor the power state of the machine or the hard drive status for HD failures.

Here is the output from ipmitool

Pwr Unit Status | 0x00 | ok
HDD 0 Status | 0x00 | ok

I added the sensor the same way I do voltage and temperature but I end up with "sensor or control Pwr Unit Status@[]:623 does not exist"

Has anyone successfully been able to monitor discrete sensors over IPMI?

Any help is greatly appreciated,

Thanks.

BTW I am on Zabbix 3

[andris]

The first thing is to get the right sensor name into Zabbix. Sensor name as discovered by OpenIPMI is similar, but not necessarily exactly the same as shown by 'ipmitools'.

https://www.zabbix.com/documentation...screte_sensors shows how one can discover sensor names using a Zabbix server with DebugLevel=4. Alternatively, OpenIPMI command line tool can be used get sensor name.
This can help to solve "sensor or control ... does not exist" problem.

[jrosetto]

Found that if I add a 0 to the end of the IPMI sensor it works. Not sure why...

My last question.

I have an Intel S2600CP4. Where would I find a list of what the IPMI codes mean to be able to setup the triggers?

[andris]

Found that if I add a 0 to the end of the IPMI sensor it works. Not sure why...

For some reason that particular OpenIPMI library version appends 0 to the name.
https://www.zabbix.com/documentation..._2018_and_2019 describe what can happen when OpenIPMI version is later upgraded.

I have an Intel S2600CP4. Where would I find a list of what the IPMI codes mean to be able to setup the triggers?

Standardized codes can be decoded using IPMI specs (can be found online) as illustrated in https://www.zabbix.com/documentation...screte_sensors.
If for example, reading_type:0x70 "OEM Discrete. Indicates that the discrete state information is specific to the OEM identified by the Manufacturer ID for the IPM device that is providing access to the sensor." or type:0xc0 "OEM RESERVED" I think only Intel can tell you what sensors they have built in there and how to decode them.

[jrosetto]

For some reason that particular OpenIPMI library version appends 0 to the name.
https://www.zabbix.com/documentation..._2018_and_2019 describe what can happen when OpenIPMI version is later upgraded.


Standardized codes can be decoded using IPMI specs (can be found online) as illustrated in https://www.zabbix.com/documentation...screte_sensors.
If for example, reading_type:0x70 "OEM Discrete. Indicates that the discrete state information is specific to the OEM identified by the Manufacturer ID for the IPM device that is providing access to the sensor." or type:0xc0 "OEM RESERVED" I think only Intel can tell you what sensors they have built in there and how to decode them.

Still trying to make sense of this...

My IPMI output looks like this

HDD 0 Status | 0x0 | discrete | 0x0100| na | na | na | na | na | na
HDD 1 Status | 0x0 | discrete | 0x0100| na | na | na | na | na | na
HDD 2 Status | 0x0 | discrete | 0x0100| na | na | na | na | na | na
HDD 3 Status | 0x0 | discrete | 0x0100| na | na | na | na | na | na
HDD 4 Status | 0x0 | discrete | 0x0000| na | na | na | na | na | na
HDD 5 Status | 0x0 | discrete | 0x0000| na | na | na | na | na | na
HDD 6 Status | 0x0 | discrete | 0x0000| na | na | na | na | na | na
HDD 7 Status | 0x0 | discrete | 0x0000| na | na | na | na | na | na

I have 4 healthy hard drives in the machine.

I am attaching a snipit from the Intel manual on what the sensor states are and if I'm reading it correctly it says they are 0x0100 which is Drive Fault. Can someone shed some light on this as I am stumped.