Oh. I forgot. I also had a problem a few weeks ago where DNS would fail completely, and NextDNS would say something in the log like too many requests at one time, or something similar. The issue is due to how NextDNS does TTL. It forces a super low (default 5 seconds) TTL on clients so they are forced to constantly query the cache on the router. This is done so changes made on your NextDNS profile (whitelisting a domain, for example) go out to clients quickly.
To fix this, I just raised the max-ttl value in the CLI configuration. Something more sensible like 30 seconds should do the trick. Or, as I've been doing lately, just set the value to 0, so clients are given the real TTL time (useful if you rarely find yourself whitelisting domains) and rely on each device's own DNS cache.