DFS provides three types of system monitoring. The first type of monitoring involves the BOS Server, which continually monitors and restarts (as necessary) all indicated DFS server processes running on a server machine. The system administrator indicates the processes the BOS Server is to monitor. The BOS Server restarts itself and all other indicated server processes on the machine once a week (to use new binary files, for example); it also checks each specified process once a day to ensure that each process is using the most current binaries. Once it is running on a machine, the BOS Server requires little intervention from the system administrator.
The second type of monitoring involves the scout program, which system administrators can use to monitor File Server machine usage. This program allows administrators to determine which machines and aggregates are experiencing the most data requests, as well as which machines are functioning properly. An administrator can use the scout program to track the File Exporters on many File Server machines from a single client machine. The scout program presents statistics about the File Server machines and File Exporters it is monitoring in a graphical format, and it allows the system administrator to set attention thresholds for the statistics being monitored.
The third type of monitoring involves the dfstrace command suite, which system administrators can use to trace DFS processes that run in either the user-space or the kernel. Commands in the suite allow sophisticated administrators and system developers to obtain internal tracing information that they can use to diagnose and debug system problems. Because dfstrace commands are beyond the scope of normal DFS administration, information about the commands is presented only in Monitoring and Tracing Tools.