Prometheus and Victoriametrics limits the scrape timeout to be equal or less than scrape interval. Reducing the defaults to 55 should prevent scrape timeouts. When using statsd use --interval 60 to ensure the entire period is covered. Ref: https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1281