In recent years, information systems have become increasingly complex, and the transmission and propagation mechanism of faults has become difficult to track, which has seriously affected the stability and reliability of the system. Therefore, an effective method is urgently needed to monitor and analyze the fault transmission of information systems. To this end, this study proposes a systematic fault transmission mechanism research method based on full-link monitoring indicators. First, a full-link monitoring framework is built to collect and analyze the monitoring data of each node to describe the path and characteristics of fault propagation. Secondly, a graph model is used to model the dependencies between services in the system, and a Transformer-based deep learning model is used to perform root cause analysis of faults. Then, the model is used to predict fault propagation trends, thereby identifying potential fault points within the system. Finally, a fault alarm and location mechanism is established to ensure that faults can be discovered and handled in a timely manner. Experimental results show that compared with long short-term memory (LSTM) and random forest models, the Transformer model has an average fault location accuracy of 94.3%, an average alarm response time of 10.3 ms, and a fault propagation path prediction accuracy. The average prediction accuracy is 92.4%, with excellent performance in all aspects, significantly shortening the system recovery time and effectively improving the system's stability and fault recovery capabilities.