Monitoring hardware metrics of physical HP systems

[m_doornbusch]

For a customer we have the requirement to monitor physicals HP servers and act on health errors of the hardware.  It’s about health of disks, motherboard, power supplies, batteries etc. 

On the server it self Windows (all kind of versions) is the operating system and a Dynatrace OneAgent active on the servers.

We don't have HPE OneView installed on the hardware.

[Peter_Youssef]

Hello @m_doornbusch 

• You can suggest installing the attached extension to the customer: remote-unix-monitoring-hp 
• For the extension ensure the general installation prerequisites are fulfilled such firewall, network connectivity ,..etc and then start adding the intended servers.
• If there is applicability to install oneagent, proceed with oneagent deployments.
• If the license consumption doesn't matter proceed with OA deployments, otherwise install the extension.

Regards,

Peter

[KevinW]

SNMP traps may be an option for you - forward traps from the ILO (HPE physical server console), or Blade enclosure console to the Environment ActiveGate.

Downside with this approach is then identifying the SNMP OID values that you need to create a problem notification for in Dynatrace and there may be specific OIDs on specific hardware types that you may need.

The ILO itself can use Network Availability Monitor to confirm its "reachable", or a HTTPS check if you want more detailed check that the ILO is alive.

HPE servers do have an on-host hardware agent (provided by HPE) that pulls hardware issues into the system log but this only works if the OneAgent is functioning. You need to identify the rules to extract the log events from the system log to create the Dynatrace Problem notifications. These rules may be different by hardware type and operating system.

Or a combination of the two above approaches.

If the hardware does not have a oneAgent, but allows a syslog integration you can use this extension: Syslog ingestion with ActiveGate — Dynatrace Docs

Additional DDU and/or DEM license consumption will occur with the above approaches as you are ingesting more data.

[PacoPorro · Accepted solution]

If SNMP is not an option for you,  you can try with ILO cmdlets + oneagent data ingest. 

https://github.com/HewlettPackard/PowerShell-ProLiant-SDK/tree/master/HPEOA/2.0
https://docs.dynatrace.com/docs/shortlink/pipe-metric-ingestion

[GilesDay]

 we'd also like to see more support for physical hardware monitoring - similar to the DellIDRAC extension

ie HPOneView with Redfish https://github.com/HewlettPackard/oneview-redfish-toolkit (it looks like the Dell iDrac extension uses Redfish)