The maximum retransmission timeout, in milliseconds, permitted by this device's TCP implementation, again with algorithm-dependent precise semantics. This ceiling controls how long the gateway's own TCP-based sessions, such as a log-forwarding connection to a remote log server, will keep backing off before giving up on a retransmission, affecting how quickly a genuinely dead peer connection is detected versus tolerated through transient loss. Example: tcpRtoMax set very high could mean a broken log-server connection takes minutes rather than seconds to be recognized as failed and retried.