In looking at the results from test1-4 AWS SSD on develop from the weekend, several tests had multi-second latencies. In particular, test4, the read-only test, had a 7+ second latency. Even though the config has max_latency=2000 it didn't stop due to a unit conversion bug that I fixed today in changeset 9ea7a82d112a8760c25f5f8da6f584f6c664cb89.

On the AWS HD system, I am seeing multi-second latencies (4+ seconds) on test1-50m, the populate test. I'm using fedorova pintools to see if I can get a trace of where that is happening. Since it is on populate, I don't think it is necessarily the same as WT-1030.