Abstract: Dynamic reconfiguration of a quorum group of processors and recovery procedure therefore are provided for a shared nothing distributed computing system. Dynamic reconfiguration proceeds notwithstanding unavailability of at least one processor of the quorum group of processors assuming that a quorum of the remaining processors exists. Recovery processing is implementing by the group of processors so that the at least one processor which was unavailable during the dynamic reconfiguration of the group is able to obtain current state information once becoming active. Each processor of the group of processors includes an incarnation number and a list of member processors which participated in a commit process resulting in its incarnation number. The recovery processing includes exchanging the processors' incarnation numbers and lists of processors for propagation of the current state of the quorum group of processors to the at least one processor now becoming available.

Abstract: A physical clock is expanded to enhance its precision. Existing instructions are capable of using the enhanced physical clock. Execution of an instruction begins, which places a value of the expanded physical clock in a physical clock field of a clock representation. The physical clock field is, however, unable to accommodate the value provided by the expanded physical clock. Thus, that value encroaches upon another predefined field of the clock representation. Completion of the instruction is therefore delayed such that the value provided by the expanded physical clock can be accommodated in the clock representation and a correct value for the another predefined field can be provided.

Abstract: A management facility for managing concurrently executable computer processes. The management facility includes a registration mechanism, which enables a plurality of interdependent processes to be considered a part of a topology of logically dependent execution groups, and a termination synchronization mechanism that synchronizes the completion of termination for the processes within the topology. Termination synchronization prevents a process within the topology from completely terminating, even if it has entered normal termination, until all of the processes within the topology have normally terminated. If one of the processes within the topology has abnormally terminated, then notification of this abnormal termination can be propagated to all of the processes within the topology of logically dependent execution groups, since none of them have completely terminated.

Abstract: Dynamic reconfiguration of a quorum group of processors and recovery procedure therefore are provided for a shared nothing distributed computing system. Dynamic reconfiguration proceeds notwithstanding unavailability of at least one processor of the quorum group of processors assuming that a quorum of the remaining processors exists. Recovery processing is implementing by the group of processors so that the at least one processor which was unavailable during the dynamic reconfiguration of the group is able to obtain current state information once becoming active. Each processor of the group of processors includes an incarnation number and a list of member processors which resulted from a commit process resulting in its incarnation number. The recovery processing includes exchanging the processors' incarnation numbers and lists of processors for propagation of the current state of the quorum group of processors to the at least one processor now becoming available.

Abstract: A program product for a message processing system in which messages are transmitted from source nodes to destination nodes. A transmission flow control technique is disclosed in which the source node optimistically sends control information and a data portion of a message, and wherein a destination node discards the data portion of the message if it is unable to accommodate it. The destination node, however, retains enough of the control information to identify the message to the source node, and when the destination node is subsequently able to accommodate the data portion, the destination node issues a request to the source node to retransmit the data portion of the message. Discarding of one message is followed by discards of sequential messages, until the destination node is able to accommodate the data portions of messages.

Abstract: Controlling the flow of information between senders and receivers across links being used as channels. In one example, a self-timed interface link is adapted to be used as a channel. Such an interface is referred to as an integrated cluster bus. The flow control for the integrated cluster bus includes, for instance, a Data Request packet that indicates to the transmitter of data that it can now send the data; a continue indicator that specifies that more data is to follow; and a sequence indicator that is used to determine if a particular message is in proper sequence order. The integrated cluster bus does not require large data buffers and offers low latency messaging.

Abstract: An initialization facility is provided for a fiber optic data link wherein the link can be initialized in one of a plurality of fiber control modes, including one or more open fiber control (OFC) modes and a non-OFC mode. Initialization may be automatic, being responsive to received optical signals, or user determined. Along with the initialization facility, safety logic is presented which allows the facility to be fully compliant with industry standard laser safety requirements for inherently safe Class 1 operation. Further, logic is provided for controlling resetting of an OFC chip. In one embodiment, the facility can be implemented in an auto-sense circuit coupled between a conventional OFC chip and a transceiver of the fiber optic data link.

Abstract: A mode conditioning patch (MCP) is provided for conditioning a single mode fiber optic signal for propagation within a multimode optical fiber. The MCP includes a single mode optical fiber section and a transition region with increasing diameter. The transition region diameter preferably increases to approximately that of a multimode fiber section through which the conditioned signal is to propagate. In one embodiment, the transition region is a flared end of the single mode optical fiber section, which is fabricated of plastic. The single mode optical fiber section with the flared end can be fusion splice joined to the multimode optical fiber section to further enhance mixing of the single mode signal within the multimode fiber optic section.

Abstract: Message-passing capability is provided in a computer system with a plurality of asynchronous computing nodes interconnected for transmission of messages between threaded user tasks executing in ones of the computing nodes. A message is received at a receiver computing node employing a threaded message passing interface (MPI), which includes a means by which a user-defined program can be called by an interrupt service thread at the MPI. The user-defined program takes a predefined action in response to the asynchronous arrival of the at least one message packet. For example, the user-defined program might comprise a program to initiate a function to receive the at least one message packet at the receiver's threaded MPI, which may include awaking a waiting thread.

Abstract: Message-passing capability is provided in a computer system with a plurality of asynchronous computing nodes interconnected for transmission of messages between threaded user tasks executing in ones of the computing nodes. A message is received at a receiver computing node employing a threaded message passing interface (MPI), which includes a means by which a user-defined program can be called by an interrupt service thread at the MPI. The user-defined program takes a predefined action in response to the asynchronous arrival of the at least one message packet. For example, the user-defined program might comprise a program to initiate a function to receive the at least one message packet at the receiver's threaded MPI, which may include awaking a waiting thread.

Abstract: A checkpoint of a parallel program is taken in order to provide a consistent state of the program in the event the program is to be restarted. Each process of the parallel program is responsible for taking its own checkpoint, however, the timing of when the checkpoint is to be taken by each process is the responsibility of a coordinating process. During the checkpointing, various data is written to a checkpoint file. This data includes, for instance, in-transit message data, a data section, file offsets, signal state, executable information, stack contents and register contents. The checkpoint file can be stored either in local or global storage. When it is stored in global storage, migration of the program is facilitated. When a parallel program is to be restarted, each process of the program initiates its own restart. The restart logic restores the process to the state at which the checkpoint was taken.

Abstract: A fiberoptic communications method is presented which allows an open-link condition on one link of a multi-link fiberoptic connection to be propagated to other links of the connection while satisfying the Open Fiber Control safety standard. Each link comprises a duplex link and at least one optical repeater is employed. The at least one optical repeater is adapted to propagate an open-link condition in one link of the multiple duplex links between the links of the fiberoptic connection. Propagating the open-link condition can be accomplished using an out band signal, an electrical wrap mode, or a hybrid approach using both an out band signal and electrical wrap mode. Automatic propagation of a closed-link condition is provided once the originally opened link is closed.

Abstract: A fiberoptic communication system is presented which allows an open-link condition on one link of a multi-link fiberoptic connection to be propagated to other links of the connection while satisfying the Open Fiber Control safety standard. Each link comprises a duplex link and at least one optical repeater is employed. The at least one optical repeater is adapted to propagate an open-link condition in one link of the multiple duplex links between the links of the fiberoptic connection. Propagating the open-link condition can be accomplished using an out band signal, an electrical wrap mode, or a hybrid approach using both an out band signal and electrical wrap mode. Automatic propagation of a closed-link condition is provided once the originally opened link is closed.

Abstract: A fiberoptic communications computer program product is presented which allows an open-link condition on one link of a multi-link fiberoptic connection to be propagated to other links of the connection while satisfying the Open Fiber Control safety standard. Each link comprises a duplex link and at least one optical repeater is employed. The at least one optical repeater is adapted to propagate an open-link condition in one link of the multiple duplex links between the links of the fiberoptic connection. Propagating the open-link condition can be accomplished using an out band signal, an electrical wrap mode, or a hybrid approach using both an out band signal and electrical wrap mode. Automatic propagation of a closed-link condition is provided once the originally opened link is closed.

Abstract: A run-time customization capability extends functionality of a software application in a computer system. Through object-oriented design, an instance of a first class is instantiated. The first class (e.g., a derived class) has a same interface as a second class. The first and second classes enable respective first and second functionalities through respective first and second implementations of the same interface. The first implementation is dynamically loaded at run time. The dynamic loading can involve locating the first implementation, such as by using a locator to locate a module comprising the first implementation. A transfer vector usable in accessing the first implementation can be initialized to have an indication of a location of the first implementation. Programming code associated with the same interface can be compiled prior to compilation of programming code associated with the first implementation.

Abstract: A system for a message processing system in which messages are transmitted from source nodes to destination nodes. A transmission flow control technique is disclosed in which the source node optimistically sends control information and a data portion of a message, and wherein a destination node discards the data portion of the message if it is unable to accommodate it. The destination node, however, retains enough of the control information to identify the message to the source node, and when the destination node is subsequently able to accommodate the data portion, the destination node issues a request to the source node to retransmit the data portion of the message. Discarding of one message is followed by discards of sequential messages, until the destination node is able to accommodate the data portions of messages.

Abstract: A checkpoint of a parallel program is taken in order to provide a consistent state of the program in the event the program is to be restarted. Each process of the parallel program is responsible for taking its own checkpoint, however, the timing of when the checkpoint is to be taken by each process is the responsibility of a coordinating process. During the checkpointing, various data is written to a checkpoint file. This data includes, for instance, in-transit message data, a data section, file offsets, signal state, executable information, stack contents and register contents. The checkpoint file can be stored either in local or global storage. When it is stored in global storage, migration of the program is facilitated. When a parallel program is to be restarted, each process of the program initiates its own restart. The restart logic restores the process to the state at which the checkpoint was taken.

Abstract: In a mainframe class data processing system having multiple logical partitions and a port to a network, a host-network interface is established for reducing network overhead at the multiple partitions. The host-network interface includes, for example, a host channel connection coupling the multiple partitions of the host system to a communications adapter having a network device driver for each network coupled to the adapter. The adapter also includes an address resolution protocol (ARP) cache designed to hold predetermined media headers for the clients coupled to the network(s) for use in forwarding an internet protocol (IP) datagram across the network to one of the clients from a partition of the host system. If not predetermined, the adapter is provided with a capability to dynamically determine the needed media header for transmission of the IP datagram across the network to the client. In Ethernet format, this dynamically determining includes forwarding an Ethernet 802.

Abstract: A checkpoint of a process is taken in order to provide a consistent state of the process in the event the process is to be restarted. When the process is restarted, the process is placed in the state it was in when the checkpoint was taken. However, there are times when certain information has changed since the last checkpoint and it should not be restored. For example, if a process is restarted on a computing unit different from the one in which the checkpoint was taken, then environment variables associated with the new computing unit, such as the name of the computing unit, are not to be restored to the values they were before the checkpoint. In order to prevent this, one or more environment variables are adjusted, after the process is restored to the reflect the state they were in prior to restoration.

Abstract: A complete and consistent set of checkpoint files is captured and identified for use in restarting a parallel program. When each process of a parallel program takes a checkpoint, it creates a checkpoint file. The checkpoint file is named, and a part of that name includes a version number for the checkpoint file. When a parallel program is to be restarted, each process of the parallel program identifies its most current valid checkpoint file. It provides the version number of this file to a coordinating process. The coordinating process then decides which version of the checkpoint files is valid and complete for all of the processes of the parallel program participating in restart. Once this version number is determined, it is forwarded to the processes and the processes restore themselves using its corresponding checkpoint file having that particular version number.