Abstract: A method, computer program product and computer system achieves full-mesh connectivity between any two domains in a multi-domain computing environment such as an Infiniband or Converged Ethernet environment. A connection between two domains is established using a single connection management identifier, and private payloads in connection management datagrams, to drive full-duplex connectivity over a pair of half-duplex connections. The half-duplex connections are established using one connection request, and one connection ID object. A connection management object interfaces between the two connected domains. The connection management object handles communications across the half-duplex connections while the connected applications operate as if they are communicating over a full-duplex connection.

Abstract: A method, computer program product and computer system achieves full-mesh connectivity between any two domains in a multi-domain computing environment such as an Infiniband or Converged Ethernet environment. A connection between two domains is established using a single connection management identifier, and private payloads in connection management datagrams, to drive full-duplex connectivity over a pair of half-duplex connections. The half-duplex connections are established using one connection request, and one connection ID object. A connection management object interfaces between the two connected domains. The connection management object handles communications across the half-duplex connections while the connected applications operate as if they are communicating over a full-duplex connection.

Abstract: Methods, computing systems and computer program products implement embodiments of the present invention that include identifying a first number of processors in a computer, and identifying a second number of interrupt request (IRQ) lines on a hardware acceleration device in the computer and coupled to the processors, the second number greater than or equal to the first number. Each of the IRQ lines is associated with one of the processors, and upon selecting a given IRQ line for an application thread, a given processor associated with the given IRQ line is identified. Execution of the application thread is initiated on the given processor, and using the given IRQ line, a completion queue is configured for the application thread. If the thread is executing on a different processor than the one managing the completion queue, then the management of the completion queue can be migrated to the processor executing the thread.

Abstract: Methods, apparatus and computer program products implement embodiments of the present invention that enable a computer system comprising networked computers to self-heal from a boot failure of one of the computers. In some embodiments, upon detecting a first computer having a memory and a physical boot device failing to successfully load a boot image, a second computer defines a virtual boot device in communication with the first computer and the second computer and having a boot image for the first computer. Subsequent to defining the virtual boot device, the second computer can configure the first computer to boot from the virtual boot device.

Abstract: A method, computer program product and computer system achieves full-mesh connectivity between any two domains in a multi-domain computing environment such as an Infiniband or Converged Ethernet environment. A connection between two domains is established using a single connection management identifier, and private payloads in connection management datagrams, to drive full-duplex connectivity over a pair of half-duplex connections. The half-duplex connections are established using one connection request, and one connection ID object. A connection management object interfaces between the two connected domains. The connection management object handles communications across the half-duplex connections while the connected applications operate as if they are communicating over a full-duplex connection.

Abstract: A method, computer program product and computer system achieves full-mesh connectivity between any two domains in a multi-domain computing environment such as an Infiniband or Converged Ethernet environment. A connection between two domains is established using a single connection management identifier, and private payloads in connection management datagrams, to drive full-duplex connectivity over a pair of half-duplex connections. The half-duplex connections are established using one connection request, and one connection ID object. A connection management object interfaces between the two connected domains. The connection management object handles communications across the half-duplex connections while the connected applications operate as if they are communicating over a full-duplex connection.

Abstract: Apparatus and computer program products implement embodiments of the present invention that enable a computer system comprising networked computers to self-heal from a boot failure of one of the computers. In some embodiments, upon detecting a first computer having a memory and a physical boot device failing to successfully load a boot image, a second computer defines a virtual boot device in communication with the first computer and the second computer and having a boot image for the first computer. Subsequent to defining the virtual boot device, the second computer can configure the first computer to boot from the virtual boot device.

Abstract: Methods, apparatus and computer program products implement embodiments of the present invention that enable a computer system comprising networked computers to self-heal from a boot failure of one of the computers. In some embodiments, upon detecting a first computer having a memory and a physical boot device failing to successfully load a boot image, a second computer defines a virtual boot device in communication with the first computer and the second computer and having a boot image for the first computer. Subsequent to defining the virtual boot device, the second computer can configure the first computer to boot from the virtual boot device.

Abstract: Provided are a method, a system, and a computer program product in which a switchless network is maintained. A processor discovers a plurality of nodes in the switchless network by requesting and receiving port information and node information in association with an attribute that allows datagrams to be propagated among the plurality of nodes of the switchless network without being terminated at host channel adapters of the plurality of nodes.

Abstract: A processor initiates an execution of a network driver in a first node comprising a master node of a plurality of nodes in a switchless network. The network driver configures the plurality of nodes and service level in the switchless network by transmitting management datagrams from the master node, wherein the management datagrams include an attribute that allows the management datagrams to be propagated among the plurality of nodes of the switchless network without being terminated at host channel adapters of the plurality of nodes.

Abstract: A plurality of nodes are maintained in a switchless network, wherein each of the plurality of nodes has at least one host channel adapter for communication. A management datagram that indicates source to destination routes for communication in the plurality of nodes is provided. An attribute is configured in the management datagram to allow the management datagram to be propagated among the plurality of nodes of the switchless network without being terminated at host channel adapters.

Abstract: A connection management message that uses a proxy attribute is received, wherein the connection management message includes information on a first proxy queue pair and a second proxy queue pair, wherein the first proxy queue pair provides communication between a proxy node and an initiator node in a switchless network, and wherein the second proxy queue pair provides communication between the proxy node and a target node in the switchless network. The connection management message that uses the proxy attribute, channels datagrams received from the initiator node to the target node in the switchless network.

Abstract: Remotely debugging a non-responsive operating system (OS) of a computer system. Central processing units (CPUs) in a computer system are bound to receive queues of a network adapter. Interrupts for a CPU is disabled, wherein the CPU is not available to process hardware interrupt requests queued in the bound receive queues. A debugging message including debugging commands is received by the network adapter, wherein the debugging message is stored in a first receive queue of the network adapter bound to a first CPU. If the first CPU is available, the debugging commands in the debugging message stored in the first of the one or more receive queues of the network adapter are identified by a debugger of the computer system. The identified debugging commands are executed by the CPU to debug the non-responsive OS of the computer system.

Abstract: Methods, computing systems and computer program products implement embodiments of the present invention that include identifying a first number of processors in a computer, and identifying a second number of interrupt request (IRQ) lines on a hardware acceleration device in the computer and coupled to the processors, the second number greater than or equal to the first number. Each of the IRQ lines is associated with one of the processors, and upon selecting a given IRQ line for an application thread, a given processor associated with the given IRQ line is identified. Execution of the application thread is initiated on the given processor, and using the given IRQ line, a completion queue is configured for the application thread. If the thread is executing on a different processor than the one managing the completion queue, then the management of the completion queue can be migrated to the processor executing the thread.

Abstract: Embodiments of the present invention are provided that include executing, by a processor, a software stack received from a first boot image, and retrieving and executing, by the processor, a second software stack. A writeable boot device such as a storage device with a removable medium is detected, and the second software stack is saved by replacing, on the writeable boot device, the first boot image with a second boot image comprising the second software stack. The second software stack is saved upon detecting the boot device having no boot image.

Abstract: In an approach for election of a manger for a network, a processor detects that a first set of management tasks is not running. A processor identifies at least a first node in a network and information about the first node. A processor determines that the first node is capable of performing the first set of management tasks based on the information about the first node. A processor elects the first node as a manager in the network to perform the first set of management tasks.

Abstract: Various embodiments of the present invention that include executing, by a processor, a software stack. A writeable boot device such as a storage device with a removable medium is detected, and a second software stack is saved by replacing, on the writeable boot device, a first boot image with a second boot image comprising a second software stack.

Abstract: Communication of a node is maintained in a switchless network via point-to-point connections of a plurality of nodes having a plurality of ports. Address Resolution Protocol (ARP) announces are performed, via a virtual network device of the node, periodically or in response to an occurrence of a change in the node, for communications among the plurality of nodes in the switchless network.

Abstract: Provided are a computer program product, system, and method for upgrading a kernel or kernel module with a configured persistent memory unused by the kernel. A persistent memory space is configured in the memory to store application data from applications in user mode. A kernel executing in the memory is prevented from accessing the persistent memory space. A service is called to load an updated kernel in the memory to replace the kernel, wherein the applications have access to the persistent memory space after the updated kernel is loaded.

Abstract: In an approach for election of a manger for a network, a processor detects that a first set of management tasks is not running. A processor identifies at least a first node in a network and information about the first node. A processor determines that the first node is capable of performing the first set of management tasks based on the information about the first node. A processor elects the first node as a manager in the network to perform the first set of management tasks.