Browse Prior Art Database

Method to Improve Safety During a Concurrent Maintenance Operation

IP.com Disclosure Number: IPCOM000173432D
Original Publication Date: 2008-Aug-05
Included in the Prior Art Database: 2008-Aug-05
Document File: 1 page(s) / 25K

Publishing Venue

IBM

Abstract

Concurrent maintenance (CM) is a feature on many mid-range and high-end systems. CM allows the system to be serviced while it is running. This maximizes uptime for the customer. However, since a running system is being serviced, there is a risk of exposure to high currents and/or voltages. Normally, the firmware shuts off the power to the part being serviced and indicates it is safe using a message passed to the management console. But, things can go wrong with the code and cause the power not to be shut off.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 87% of the total text.

Page 1 of 1

Method to Improve Safety During a Concurrent Maintenance Operation

     Proposed is a method to read the currents and voltages directly from the power supply. This way, the service person does not have to rely on software and message passing for their personal safety. If all voltages and currents are 0, the user will know it is safe. Currently, the firmware doesn't have any way to check to see if the voltages and currents are off for that specific node. During boot, it sends the power on command to the other nodes and assumes the failed node is not powered on. Things can go wrong in code, and on the hardware that could cause the node to be powered on or never powered off.

     The power supplies on high-end systems regulate voltages using voltage regulator modules (VRMs). Each VRM supplies a separate voltage plane. The power supplies are sophisticated and have code to read the current and voltage for each VRM. The power supplies interface with the bulk power controller (BPC) via serial cables. The BPC is a service processor running Linux* or similar operating system. The system management console, such as an HMC, can interface with the BPC using the existing ethernet path to query the voltage and current.

     The user would initiate the concurrent maintenance procedure, then the HMC would send a query voltage and current command to the BPC. The BPC would pass the query on to the appropriate power supply and receive the data back from it...