According to the HAProxy documentation the metric qtime represents the average queue time over the last 1024 requests. My question is, does this metric average ALL requests, or just those that have sat in the queue?
For example, if I have received 1000 requests, and 10 of those were queue, does qtime represent the total time spent in queue averaged over all 1000 requests or only the 10 that were actually queued?
