When we were calculating the start time for a request, we were checking to see if start > 0. That meant that on the initial request, if the start time was 0 exactly, we were incorrectly calculating the all_start and all_end times, effectively leaving the DNS timings out.
When we were calculating the start time for a request, we were checking to see if start > 0. That meant that on the initial request, if the start time was 0 exactly, we were incorrectly calculating the all_start and all_end times, effectively leaving the DNS timings out.
This fixes that by checking if >= 0 instead.