We've noticed that there is a higher-than-expected amount of IP churn for our wireless clients. That is, clients are connecting to the same network, but being assigned different IP addresses in the pool.<div><br></div>
<div>I don't believe it's due to over-subscription. In the past few months, we've had a total of 13K unique devices with 12K available IP addresses. This might affect a transient device or two, but our long-term customers should retain the same IP.</div>
<div><br></div><div>We do; however, have shared-networks setup. Each shared-network consists of a /21 (2000 clients) and a /23 (500 clients).</div><div><br></div><div>So here's the question ... when there are two subnets defined inside a shared-network, does dhcpd always prefer using A, even if the client has an existing lease from B? That is, does it only use B if A is completely filled with active clients? This would certainly explain the behavior we're observing.</div>
<div><br></div><div>Thanks!</div><div><br></div><div>Norman Elton</div><div><br></div>