This paper proposes a bio-inspired service provisioning scheme for fog-based industrial Internet of Things (IIoT). We use wild-goats algorithm to address the adaptive fog service provisioning problem. Specifically, the master node is configured to monitor the network condition of all fog nodes inside of its domain. Then, the master node will distribute a service to the most suitable fog node based on network latency, computation cost, and storage cost. Therefore, the service can be accomplished in a real-time. The proposed method also considers a probability of node failure during the service provisioning process due to the harsh condition of IIoT environment. The results show that the proposed scheme can achieve low latency compared to the traditional architectures. The results also show that the node failure can escalate the latency of fog-based IIoT network. Thus, the adaptive service provisioning scheme is essential to maintain the network performance of fog-based IIoT.