This paper is concerned with the analysis and redesign of a distributed algorithm to monitor the availability of nodes in self-configuring networks. The simple scheme to regularly probe a node — "are you still there?" — may easily lead to over- or underloading. The essence of the algorithm is therefore to automatically adapt the probing frequency. We show that a self-adaptive scheme to control the probe load, originally proposed as an extension to the UPnPTM (Universal Plug and Play) standard, leads to an unfair treatment of nodes: some nodes probe fast while others almost starve. An alternative distributed algorithm is proposed that overcomes this problem and that tolerates highly dynamic network topology changes. The algorithm is very simple and can be implemented on large networks of small computing devices such as mobile phones, PDAs, and so on.