Configuration debugging as search: finding the needle in the haystack

摘要: This work addresses the problem of diagnosing configuration errors that cause a system to function incorrectly. For example, change local firewall policy could network-based application malfunction. Our approach is based on searching across time for instant transitioned into failed state. Based this information, troubleshooter or administrator can deduce failure by comparing state before and after failure. We present Chronus tool, which automates task failure-inducing change. takes as input user-provided software probe, differentiates between working nonworking states. performs "time travel" booting virtual machine off system's disk it existed at some point in past. By using binary search, find fault with effort grows logarithmically log size. We demonstrate diagnose range common both client-side server-side applications, performance overhead tool not prohibitive.