Server Reintegration in a Replicated UNIX File System

A distributed system is a collection of nodes connected by a network, an ideal platform to provide high reliable computing due to the redundancy supplied by a great number of nodes. Node faults and network connection faults can be masked recon guring the system. However, sequential faults, that a ect multiple nodes can decrease the performance of the system a ecting the system reliability and availability. To avoid this, failed nodes should be reintegrated as soon as possible. This paper details the problem of reintegration of failed nodes in a replicated UNIX le system. We built a prototype with the recovery protocols required by the reintegration procedure.