Gossip network protocol is widely used as the underlying communication protocol of the blockchain distributed system due to its high efficiency and scalability.Since the propagation efficiency is reduced by the redundancy of Gossip network protocol in the alliance chain,Stack Gossip algorithm is proposed.The information of the node receiving the message is recorded in the information stack by the algorithm,in order to avoid repeatedly sending the message to one node.During the experiment,it is found that Stack Gossip algorithm is not suitable for the large-scale network of too many nodes,so Influence Gossip algorithm is further proposed.The core idea of Influence Gossip algorithm is that the peer node to communicate with is selected by the node in the way of evaluating the information influence of the neighbour node,avoiding the randomness of the propagation between the nodes of the traditional Gossip network protocol to some extent.The experimental results show that compared with Random Gossip algorithm,Influence Gossip algorithm and Stack Gossip algorithm have obvious improvement in the propagation efficiency and redundancy.