The invention provides a Linux cluster fault automatic recovery method which comprises the following steps of: executing data information acquisition and judging whether a fault occurs; if the fault occurs, restarting a node; after the node is re-started, executing the data information acquisition again, and if the fault occurs, executing maintenance integration of the fault node; after executing the maintenance integration of the fault node, executing the data information acquisition again, and if the fault occurs, executing installation integration of the fault node; and after executing the installation integration of the fault node, executing the data information acquisition again, and if the fault occurs, performing manual treatment. The Linux cluster fault automatic recovery method reduces the labor consumption to a great extent, can automatically, quickly and efficiently finish fault automatic recovery of a cluster node system, can meet different needs of a heterogeneous cluster, supports multiple versions of operating systems, and improves the utilization efficiency of the cluster resources.