Abstract
This thesis presents a measurement-based dependability ics. study of a UNIX server. The event logs of a UNIX server are collected to form a dependability data basis. Message logs spanning approximately eleven months were collected for this study. The event log data are classified and categorized to calculate parameters such as MTBF and availability to characterize the dependability of the system. Component analysis is also performed to identify server modules that are prone to errors in the system. Next, the system error activity preceding each system failure is analyzed to identify error patterns which may be used to predict future failures. Lastly, the error/failure results of fault-injection experiments previously reported on computer systems are compared to the measurement data from the event logs.
Lal, Ronjeet Singh (1998). Measurement-based study of failures in a UNIX server. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1998 -THESIS -L35.