Abstract: Methods, systems and computer program products are provided for monitoring a task executing on a data processing system, the task having an associated work in process queue and an associated work pending queue. The task is configured to properly execute requests that are terminated in progress and restarted from an initial start point. A watchdog task determines if the task is executing properly and restarts the task if it is not executing properly. Restarting is provided by placing requests in the work in process queue of the terminated task in the work pending queue and clearing the work in process queue. Execution by the task of requests from the work pending queue is then reinitiated.

Abstract: Methods, systems and computer program products are provided for monitoring a task executing on a data processing system, the task having an associated work in process queue and an associated work pending queue. The task is configured to properly execute requests that are terminated in progress and restarted from an initial start point. A watchdog task determines if the task is executing properly and restarts the task if it is not executing properly. Restarting is provided by placing requests in the work in process queue of the terminated task in the work pending queue and clearing the work in process queue. Execution by the task of requests from the work pending queue is then reinitiated.

Abstract: Methods, systems and computer program products are provided which coordinate operations for a plurality of interrelated tasks executing on a computer using actual state objects and desired state objects. For each event including coordination between two of the plurality of interrelated tasks, a first (or initiator) task initiates operations by a second (or executor) task to carry out a desired sequence of operations. The initiator task sets a desired state object to the desired state and submits a request to the executor task. The executor task, in turn, operates on the request in order to update an actual state object to the desired state stored in the desired state object by the initiator task. Write control over the desired state object is therefore granted to the initiator task while write control over the actual state object is granted to the executor task.

Abstract: A method of monitoring the topology of a network uses dynamic switching between a polling mode and a event-monitoring mode. The network is of the type having point-to-point connections between a number of nodes. In the polling mode, a manager coupled to one of the nodes sends a request to an agent in each of the nodes for information about operativeness of the node, its agent, and its connections to other nodes. All of this information is collected to generate a topology map, which may be visually displayed. In the event-monitoring mode, the manager waits for event-messages from the agents in the nodes, to update the topology map. The manager switches from event monitoring mode (which is more efficient) to polling (which is usually more accurate) in response to a number of factors, including the reachablity of the nodes, the manageability of the nodes, and the consistency of information received from various nodes. Reachability is an indication of whether an enabled connection exists from manager to node.