The Jenkins job for long wtperf tests measure the maximum latency seen in each test and plots them. Since the merge of WT-4758 those latencies have been measured at essentially 0. That is very unexpected. That ticket modified wtperf itself only.

I suspect that either something in that change broke the calculation or format output of the latency information. The Jenkins configuration knows/hard-codes the CSV format column of the data it is looking for. So if that information moved, then the job should be updated.