Abstract: A method initializes a federated computer system from a fabric of nodes connected by a federated interface. Each node casts a vote to the federated interface for a candidate firmware version supported by the node casting the vote. The candidate firmware version having received the greatest number of votes is identified, and the computer system is initialized as a federated system of the nodes that support the firmware version identified as having received the greatest number of votes. A process of iterative voting may be used to identify a greater number of nodes supporting a compatible firmware version.

Abstract: A method for firmware update coordination in a federated computing system can include receiving in a host one of different computers coupled to one another in a federated computing system, different version levels of firmware versions in peer ones of the different computers. The method additionally can include recording in a compatibility list in the host computer, different version levels of firmware versions for respectively different peer ones of the computers. The method yet further can include selecting a firmware update to be applied to the host computer and comparing a version level of the selected firmware update to the different version levels in the compatibility list. Finally, the firmware update can be applied to the host computer only if the version level of the selected firmware update is present in the compatibility list, but otherwise rejecting the firmware update.

Abstract: Methods, apparatus, and products for backing up an image in a computing system that includes computer memory, including: receiving, by a backup image manager, an image for one or more computing devices within the computing system; identifying, by the backup image manager, available protected computer memory within the computing system, wherein the available protected computer memory within the computing system is restricted from alteration by a user of the computing system; slicing, by the backup image manager, the image into a plurality of image slices; and storing, by the backup image manger, one or more of the image slices in the available protected computer memory.

Abstract: In a data center that includes a number of chassis, with each chassis including a chassis management module and a number of slots, each slot associated with a service processor: network address assignment includes discovering, by a chassis management module of a particular chassis, one or more other chassis management modules; negotiating, by the chassis management module with the other chassis management modules via IPv6 data communications, a chassis-specific range of IPv4 addresses available for assignment to service processors of the particular chassis; providing, by the chassis management module to the service processors of the particular chassis, in dependence upon the chassis-specific range of IPv4 addresses, a base IPv4 address; and determining, by each service processor of the particular chassis, in dependence upon the base IPv4 address and a slot identifier of the slot associated with the service processor, an IPv4 address for the service processor.

Abstract: Data processing workload control in a data center is provided, where the data center includes computers whose operations consume power and a workload controller composed of automated computing machinery that controls the overall data processing workload in the data center. The data processing workload is composed of a plurality of specific data processing jobs, including scheduling, by the workload controller in dependence upon power performance information, the data processing jobs for execution upon the computers in the data center, the power performance information including power consumption at a plurality of power-conserving states for each computer in the data center that executes data processing jobs and dispatching by the workload controller the data processing jobs as scheduled for execution on computers in the data center.

Abstract: A computer program product includes computer usable program code embodied on a computer usable storage medium for performing a method of managing the workload in a computer system having one or more semi-redundant hardware components are provided. The method comprises detecting loss or degradation of the level of performance of one or more of the semi-redundant hardware components, identifying hardware components that are affected by the loss or degradation of the one or more semi-redundant components, migrating a critical job from an affected hardware component to an unaffected hardware component, and performing less-critical jobs on an affected hardware component. Loss or degradation of the semi-redundant component reduces the capacity of affected hardware components in the computer system without entirely disabling the computer system. Optionally, the semi-redundant hardware component may be selected from a memory module, CPU core, Ethernet port, power supply, fan, disk drive, and an input output port.

Abstract: Checkpoint snapshots of segments of system memory are taken while an operating system is booting in a computer system. The segments of system memory are stored in non-volatile memory as hibernation files. In response to detecting a request for a system reboot of the OS, an affected hibernation file, which corresponds to an affected segment of system memory that will change during the system reboot of the OS, is identified. A restoration of the system memory via a wake-up from hibernation is then initiated. The wake-up from hibernation proceeds until the affected hibernation file is reached, such that initial steps in the system reboot are bypassed. Thereafter, subsequent steps, which are after the bypassed initial steps in the system reboot, are executed.

Abstract: Embodiments include a power-efficient failover system. In one embodiment, a primary server is operable at one or more power states and is configured to dynamically generate a backup for the results of n executed program code while in a normal operating state. A redundant server is coupled to the primary server and is operable at the normal operating state or one or more reduced power states. The redundant server is also configured to dynamically receive the backup from the primary server and, in response to a failure of the primary server, to assume the workload of the primary server according to the backup. A controller manages the power state of the redundant server.

Abstract: Managing firmware in a computing system storing a plurality of different firmware images for the same firmware includes: calculating, for each firmware image in dependence upon a plurality of predefined factors, a preference score; responsive to a failure of a particular firmware image, selecting a firmware image having a highest preference score; and failing over to the selected firmware image.

Abstract: A driver is provided to manage launching of tasks at different levels of priority and within the parameters of the firmware interface. The driver includes two anchors for managing the tasks, a dispatcher and an agent. The dispatcher operates at a medium priority level and manages communication from a remote administrator. The agent functions to receive communications from the dispatcher by way of a shared data structure and to launch lower priority level tasks in respond to the communication. The shared data structure stores communications received from the dispatcher. Upon placing the communication in the shared data structure, the dispatcher sends a signal to the agent indicating that a communication is in the data structure for reading by the agent. Following reading of the communication in the data structure, the agent launches the lower priority level task and sends a signal to the data structure indicating the status of the task.

Abstract: A method of managing the workload in a computer system having one or more semi-redundant hardware components is provided. The method comprises detecting loss or degradation of the level of performance of one or more of the semi-redundant hardware components, identifying hardware components affected by the loss or degradation, migrating a critical job from an affected hardware component to an unaffected hardware component, and performing less-critical jobs on an affected hardware component. Loss or degradation of the semi-redundant component reduces the capacity of affected hardware components in the computer system without entirely disabling the computer system. Jobs identified as critical run on hardware components having the most capacity and reliability, while less-critical jobs use the remaining capacity of affected hardware components. Examples of semi-redundant hardware components include a memory module, CPU core, Ethernet port, power supply, fan, disk drive, and an input output port.

Abstract: Embodiments include a power-efficient failover method. The method includes operating a primary server at a normal operating state in which program code is executed, and dynamically generating a backup of the results of the executed program code while in the normal operating state. The method further includes operating a redundant server at a reduced power state in which less power is consumed than in the normal operating state of the primary server. The workload of the primary server may be assumed according to the backup in response to a failure of the primary server. The power state of the redundant server is managed, including maintaining the redundant server in the reduced power state prior to detecting a failure of the primary server and increasing the power state of the redundant server and assuming the workload of the primary server in response to the failure of the primary server.

Abstract: A driver is provided to manage launching of tasks at different levels of priority and within the parameters of the firmware interface. The driver includes two anchors for managing the tasks, a dispatcher and an agent. The dispatcher operates at a medium priority level and manages communication from a remote administrator. The agent functions to receive communications from the dispatcher by way of a shared data structure and to launch lower priority level tasks in respond to the communication. The shared data structure stores communications received from the dispatcher. Upon placing the communication in the shared data structure, the dispatcher sends a signal to the agent indicating that a communication is in the data structure for reading by the agent. Following reading of the communication in the data structure, the agent launches the lower priority level task and sends a signal to the data structure indicating the status of the task.

Abstract: In one embodiment, a computer system has a plurality of chassis interconnected by cables. Each cable initially connects a unique port pair consisting of a port of one chassis and a port of another chassis. The disconnection of cables is monitored, including electronically recording a disconnection sequence in which the port pairs are disconnected by removing each cable from at least one port of the respective port pair. Visual guidance is provided for re-cabling the computer system by sequentially identifying the port pairs in a reversal of the disconnection sequence, wherein identifying each port pair includes illuminating visual indicators associated with the ports of each port pair with a matching illumination pattern.

Abstract: Embodiments include a power-efficient failover system and method. In one embodiment, a primary server operating in a normal operating state is configured to dynamically backup device states or transaction logs. A redundant server coupled to the primary server in a failover cluster is operated at a reduced power state. The redundant server dynamically receives the backup from the primary server and is elevated to a normal operating state in response to a failure of the primary server. By enforcing a reduced power state of the redundant server, a failover system provides a desired combination of high power efficiency with low latency.

Abstract: Methods and systems for controlling fluid coolant flow in cooling systems of computing devices are disclosed. According to an aspect, a method may include determining a temperature of a fluid coolant in a cooling system of a computing device. For example, a temperature of water exiting a cooling system of a server may be determined. The method may also include determining an operational condition of the computing device. For example, a temperature of a processor, memory, or input/output (I/O) component may be determined. Further, the method may include controlling a flow of the fluid coolant through the cooling system based on the temperature of the fluid coolant and/or the operational condition.

Abstract: A computer program product and method of managing the workload in a computer system having one or more semi-redundant hardware components are provided. The method comprises detecting loss or degradation of the level of performance of one or more of the semi-redundant hardware components, identifying hardware components that are affected by the loss or degradation of the one or more semi-redundant components, migrating a critical job from an affected hardware component to an unaffected hardware component, and performing less-critical jobs on an affected hardware component. Loss or degradation of the semi-redundant component reduces the capacity of affected hardware components in the computer system without entirely disabling the computer system. Jobs identified as being critical are run on hardware components having the most capacity and reliability, while allowing less-critical jobs to make use of the remaining capacity of affected hardware components.

Abstract: Aspects of a distributed computing environment with remote data collection management include a management server, and a plurality of computer systems networked to and managed by the management server. Each of the plurality of computer systems includes a listening agent to receive process code from a management server and to control the periodic execution of the process code to collect data from the plurality of computer systems. The periodic execution occurs without requiring a connection to be maintained between the plurality of computer systems and the management server. Further, the data collected includes data collected from problem determination activities.

Abstract: Checkpoint snapshots of segments of system memory are taken while an operating system is booting in a computer system. The segments of system memory are stored in non-volatile memory as hibernation files. In response to detecting a request for a system reboot of the OS, an affected hibernation file, which corresponds to an affected segment of system memory that will change during the system reboot of the OS, is identified. A restoration of the system memory via a wake-up from hibernation is then initiated. The wake-up from hibernation proceeds until the affected hibernation file is reached, such that initial steps in the system reboot are bypassed. Thereafter, subsequent steps, which are after the bypassed initial steps in the system reboot, are executed.

Abstract: Data processing workload control in a data center is provided, where the data center includes computers whose operations consume power and a workload controller composed of automated computing machinery that controls the overall data processing workload in the data center. The data processing workload is composed of a plurality of specific data processing jobs, including scheduling, by the workload controller in dependence upon power performance information, the data processing jobs for execution upon the computers in the data center, the power performance information including power consumption at a plurality of power-conserving states for each computer in the data center that executes data processing jobs and dispatching by the workload controller the data processing jobs as scheduled for execution on computers in the data center.