To reduce the number of stable storage accesses and impose no restriction on the execution of live processes during recovery, Elnozahy proposed a recovery algorithm based on causal message logging. However, the algorithm with independent checkpointing may force the system to be in an inconsistent state when processes fail concurrently. In this paper, we identify these inconsistent cases and then present a recovery algorithm to perform consistent recovery by allowing the recovery leader to collect recovery information from the other recovering processes as well as all live ones. Our recovery algorithm requires no additional message compared with Elnozahy’s algorithm.