There are two assertions in the JS test for mirrored reads FTDC metrics (see here and here) that timeout after 5 seconds. On slow machines, querying FTDC metrics for the primary might take longer, resulting in failures. We should either increase the timeout or restructure the test to handle slow machines.