1. A Real Time Clock has nothing to do with RealTime scheduling. 2. I don't think that these time windows in DNSSEC are so tight that real time scheduling would make any difference. 3. At first glance I don't see any syscalls asking for RealTime scheduling in dnsmasq, so I guess enabling it in the kernel for the process would not change anything. Is there any indication that the transient failures you are seeing are caused by missed time windows?