class
Cryload::Stats
- Cryload::Stats
- Reference
- Object
Overview
Stats holder for the benchmark.
Defined in:
cryload/stats.crConstant Summary
-
PER_URL_LIMIT =
1000 -
Per-URL histograms are bounded: a cache-busting list of tens of thousands of URLs would spend more memory on the breakdown than the run is worth, and per-URL percentiles over a handful of samples each are meaningless.
-
PHASE_PERCENTILES =
[50.0, 90.0, 95.0, 99.0, 99.9] -
REPORT_PERCENTILES =
[10.0, 25.0, 50.0, 75.0, 90.0, 95.0, 99.0, 99.9]
Constructors
Instance Method Summary
- #average_bytes_per_response : Float64
- #benchmark_start : Time::Instant
- #breached_thresholds : Array(ThresholdResult)
- #bytes_per_second : Float64
- #ci_thresholds : CiThresholds
- #config : RunConfig
- #corrected_latency_view : LatencyView
- #csv_output? : Bool
- #deadline : Time::Instant | Nil
- #duration_mode? : Bool
-
#effective_latency_view : LatencyView
The histogram the CI gates read: corrected unless the user turned correction off.
- #empty? : Bool
- #error_breakdown : Array(ErrorEntry)
- #exit_code : ExitCode
- #failure_flags_breached? : Bool
- #failure_rate_percent : Float64
- #json_output? : Bool
- #latency_histogram_bins(bin_count : Int32 = 11)
- #latency_view : LatencyView
- #mark_benchmark_end(at : Time::Instant = Time.instant) : Nil
- #merge_batch(batch : Batch) : Nil
- #new_batch : Batch
- #not_ok_requests : Int64
- #ok_requests : Int64
- #output_format : String
- #phase_views : Hash(String, LatencyView)
- #progress_enabled? : Bool
- #quiet_output? : Bool
- #rate_report : RateReport
- #record(sample : Sample) : Nil
- #record_error(category : String, message : String | Nil = nil, url_index : Int32 = 0) : Nil
- #request_number : Int32
- #request_per_second : Float64
- #response_count : Int64
- #send_delay_view : LatencyView
-
#start_benchmark_window : Nil
Warmup runs before the timed window; v5 started the clock before warmup, which deflated throughput and shortened the duration window by the warmup length.
- #status_breakdown : Array(StatusEntry)
- #status_code_counts : Hash(Int32, Int64)
- #success_status_ranges : Array(Range(Int32, Int32))
- #target_unreachable? : Bool
- #text_output? : Bool
- #threshold_results : Array(ThresholdResult)
- #total_request_count : Int64
- #total_response_bytes : Int64
- #track_urls? : Bool
- #transport_error_count : Int64
- #url : String
- #url_breakdown : Array(UrlEntry) | Nil
- #urls : Array(URI)
- #wall_clock_seconds : Float64
Constructor Detail
Instance Method Detail
The histogram the CI gates read: corrected unless the user turned correction off. Identical to the service histogram without --rate.
Warmup runs before the timed window; v5 started the clock before warmup, which deflated throughput and shortened the duration window by the warmup length.