Hotplug Terminology Index

availability

A term used to indicate the probability of a system or capability of a system to remain operable and service requests. Availability takes into account the operating time of the system or system capability and the time required for repairs or maintenance. It is typically expressed in terms of the percentage of time the system remains in service (99.999% or 99.9999%, implying a service downtime of 5 minutes or 30 seconds per year). Hotplug will increase the availability of a system by allowing the replacement of components before their failure.

coldplug

Adding a component before the system is fully bootstrapped to multi-user mode.

failover

An operational mode where the function of a system component is assumed by a standby component when the primary system component becomes unavailable because of failure or scheduled down time.

hot-add

Plug in a component and expect it to be immediately usable as an integral part of the system.

hotplug

Conceptually hotplug is a means to add or remove a component while the system is running without requiring a reboot. Hotplug is a combination of both hot-add and hot-remove. In the Linux implementation of this concept, hotplug is a facility that supports dynamic (re)configuration of GNU/Linux distributions by kernel reports to user mode "agent" software.

hotplug agents

(Linux implementation specific) Scripts called by the hotplug code in response to different types of hotplug events generated by the kernel. Their action is to insert corresponding kernel modules and call any user-provided scripts.

hotplug environment variables

(Linux implementation specific) Parameter values passed to the hotplug agents by the hotplug system service to indicate what action and on what component needs to be taken.

hotplug events

(Linux implementation specific) Hotplug events are triggered by the kernel when a link is established to a device, when a driver registers an interface or by any changes made to sysfs. The hotplug events are processed by the responsible hotplug agent.

hotplug filters

(Linux implementation specific) Functions that allow the driver core to police what hotplug events are allowed.

hot-remove

Unplug a component and expected it to be taken off the resource list of the system without requiring a reboot or affecting the stability of the entire system.

node

Device that contains any possible combinations of CPUs, memory and I/O.

redundancy

Duplication of hardware or data paths which exist to prevent the failure of the active system upon failure of the duplicated component.

reliability

The probability that a system or a capability of a system will continue to function without failure under given conditions for a specified period of time.

sysfs

Memory based file system that provides a means of exporting certain attributes and information (internal kernel data structures) to userspace as normal files/symlinks in directories (replacement for procfs). It has several directory hierarchies showing the available hardware devices and modules/drivers attributes (one attribute per file). Any changes to sysfs are reflected by hotplug events generated by the kernel.

udev

Userspace component that manages the /dev device directory and creates devices based on hotplug events (replacement for devfs). Udev moves the creation and naming of the devices files to userspace.

Other Terms

Patch Lifecycle Manager (PLM)

A source manager which provides a central point for tracking the lifecycle of patches against a source tree.

Scalable Test Platform (STP)

Test harness for building reproducible test environments, executing tests and collecting results.

System Under Test (SUT)

The system which we will be executing any binary regression tests on.

use case

Hypothetical scenario as to how end users will exercise and benefit from use of hotplug capabilities.