A core virtualization host raised a hardware alert: power supply unit (PSU) number two had failed completely. The server still operated normally, for its power was redundant, yet it labored under a very heavy load.

The faulty unit had to be replaced, and replaced while the server ran, lest I be forced to migrate all the virtual machines.

In the management console I confirmed that the remaining power supply unit was providing enough power for the entire server. Then I went to the rack with a replacement unit, removed the power cable from the faulty one, unlocked the retaining clip, and drew it out of the chassis. The new unit I pushed into the empty slot until the clip locked, and connected its power cable.

The status light on the new power supply unit turned green. The server management console reported PSU number two online and functioning correctly. Redundancy was restored, and the virtual machines suffered no downtime during the replacement.