This paper analyzes the strategies for the behavior of cyber-physical systems in case of data integrity loss during transmission by nodes, temporary loss of system connectivity, and failure of individual nodes. Some problems causing the malfunction of cyber-physical systems are considered, as well as the methods, proposed for detecting and eliminating the consequences of such integrity loss. The categorization of these methods is carried out. The considered methods are divided into ontological, reorganizational ones, and methods of assessing and comparing states. When using ontological methods, databases with predetermined failure modes, implicit redundancy, and a system architecture that included several types of agents are commonly used. Most of the considered reorganizational methods are based on system reorganization with the removal of faulty nodes or their clustering with the aid of diagnostic agents. In the case of reorganization, decision-making logic, system diagnostics, and actionable scenarios for failure cases are commonly used. When using methods for comparing system states to detect failures, the following were used: categorization of nodes, controllers and monitors, hash functions, and dynamically updated keys. The results of this survey can be applied in developing of proprietary method for detecting failures and automatic recovery of cyber-physical systems.