🔴 HIGH SEV · Inference In-Log Wait Severe (p99 over 30s) 🔥 In-log wait p99 over 30s - enforcement is materially behind <|📈 graph> · <|📊 service dashboard> · :rotating_light: alarm definition · :no_bell: silence _HIGH SEV escalation of the LOW SEV 'Inference Consume Latency p99 High' (3s). At 30s p99 the consumer is far enough behind that the whole enforcement chain is pushed out. HOW THIS REACHES ENFORCEMENT (indirect but decisive): the proxy's block decision is synchronous and domain-driven - resolve domain -> activityIds, check the proxyblock{device}:{activity} Redis key - and never consults a foreground prediction at connection time. What predictions govern is whether that key is ever WRITTEN: inference attributes screen time, the limit is reached, ozone-api sets the key. So a consumer running behind delays the prediction, the billed time, the threshold, and finally the block - and the kid keeps browsing past their limit meanwhile. Prod baseline 0.06-0.46s. A single 2.05s spike appeared in 7d and one excursion into the 10-30s bucket was visible on the heatmap, but neither sustained 15m, so this should stay quiet in normal operation. no_data=OK: histogram_quantile returns NaN when the consumer is gone, and that case is covered by the coverage/starved criticals._ Grafana vmain-434316c