HPC Cluster Blog – IPMI

Posted on May 6, 2014

Nearly all of the systems shipped from Advanced Clustering in the last couple of years come standard with IPMI. IPMI stands for Intelligent Platform Management Interface, and gives you powerful remote management features for your compute nodes and servers.

IPMI runs as part of a small co-processor, also known as the BMC, or baseboard management controller. The BMC is actually integrated in the video chipset of our systems. A couple of key facts about the BMC:

  • The BMC has its own RAM, ethernet port and operating system
  • When power is applied to the system the BMC boots first, then it powers on the host system
  • It is always running as long as the power cable is inserted, even when the system is powered off
  • Has direct hardware monitoring of system sensors
  • Can re-direct the system console to view remotely

The three main features of IPMI that make remote management easy:

  1. Remote power control – Since the IPMI BMC is running even when the system is off, this gives you the ability to control the power remotely. This even includes turning a system on that is not running.
  2. Sensors – The BMC interfaces directly with the hardware sensors built into the systems. The 3 major areas that can be monitored are: Temperature, Fans, and Voltages. Thresholds can be put in place to alert and log when values are outside of normal ranges.
  3. Remote Console – There are two options for remote console support with our systems, IPMI SOL (Serial Over LAN) or iKVM. The SOL features provide a text console, similar to serial redirection, iKVM provides a full graphical redirection even with X11 or Windows apps. This requires connecting with a web-browser with Java applet support.

Cluster wide usage of IPMI
Since we specialize in delivering large scale systems with many nodes, managing IPMI one at a time would be a painful process. As part of our ACT_Utils suite, that comes free with a cluster purchase, we make using IPMI at scale much easier. There are tools like act_powerctl, act_sensors, and act_console. These allow you to control or query cluster wide, or a subset of nodes, BMCs with one simple command.

Have more questions about IPMI?
Give us a call at 866.802.8222 as we would be happy to help you understand how you can put IPMI to use on your system.

Download our HPC Pricing Guide
Get our Guide to Grant Writing

Request a Consultation from our team of HPC and AI Experts

Would you like to speak to one of our HPC or AI experts? We are here to help you. Submit your details, and we'll be in touch shortly.

  • This field is for validation purposes and should be left unchanged.