A resource manager (RM) instance is associated with each transaction processing system (TPS) member, of a TPS group. Each RM instance monitors performance of the associated TPS member. If a TPS member becomes unavailable for any reason (a failing TPS), the associated RM instance broadcasts status of the failing TPS to RMs associated “surviving” members of the group. RM instances associated with surviving members initiate a series of actions that reduce the resources used by the surviving TPS members. Consequently, the surviving TPS members are better able to process the additional workload imposed on them due to the unavailability of the failing TPS. Once the failing TPS is brought back online and made available again (or a replacement TPS is brought online), RM instances associated with the surviving members perform actions to undo the resource usage reduction tasks, and the TPS group returns to a nominal configuration.