Online tech support communities have become valuable channels for users to seek and provide solutions to specific problems. From the resource exchange perspective, the sustainability of a social system is contingent upon the size of its members as well as their communication activities. To further extend the resource-based model, the current research identifies a variety of social roles in a large tech support Q&A forum and examines longitudinal changes in the community's structure based on the identification. Moreover, this study also investigates the relationship between the community's functionality and its traffic. Results suggest that the proportion of unsolved questions negatively impacts the number of future incoming questions and the outcome of a given question is not only dependent on users' interactions within the discussion, but also on the community activities preceding the question. These observations can help community managers to improve system design and task allocation.