Uploaded image for project: 'JDK'
  1. JDK
  2. JDK-8258064

JfrGTestAdaptiveSampling needs hardening

XMLWordPrintable

    • Icon: Enhancement Enhancement
    • Resolution: Unresolved
    • Icon: P3 P3
    • tbd
    • 16, 17
    • hotspot
    • None
    • jfr

      [----------] 8 tests from JfrGTestAdaptiveSampling
      [ RUN ] JfrGTestAdaptiveSampling.uniform_rate_vm
      === Adaptive sampling: random uniform, all samples
      [ OK ] JfrGTestAdaptiveSampling.uniform_rate_vm (139 ms)
      [ RUN ] JfrGTestAdaptiveSampling.low_rate_vm
      === Adaptive sampling: low rate
      [ OK ] JfrGTestAdaptiveSampling.low_rate_vm (0 ms)
      [ RUN ] JfrGTestAdaptiveSampling.high_rate_vm
      === Adaptive sampling: high rate
      test_adaptiveSampler.cpp:231: Failure
      The difference between target_sample_size and sample_size is 10123, which exceeds expected_sample_points * error_factor, where
      target_sample_size evaluates to 500000,
      sample_size evaluates to 489877, and
      expected_sample_points * error_factor evaluates to 10000.
      Adaptive sampling: high rate
      [ FAILED ] JfrGTestAdaptiveSampling.high_rate_vm (269 ms)
      [ RUN ] JfrGTestAdaptiveSampling.early_burst_vm
      === Adaptive sampling: early burst
      [ OK ] JfrGTestAdaptiveSampling.early_burst_vm (26 ms)
      [ RUN ] JfrGTestAdaptiveSampling.mid_burst_vm
      === Adaptive sampling: mid burst
      [ OK ] JfrGTestAdaptiveSampling.mid_burst_vm (26 ms)
      [ RUN ] JfrGTestAdaptiveSampling.late_burst_vm
      === Adaptive sampling: late burst
      [ OK ] JfrGTestAdaptiveSampling.late_burst_vm (26 ms)
      [ RUN ] JfrGTestAdaptiveSampling.bursty_rate_10_percent_vm
      === Adaptive sampling: bursty 10%
      [ OK ] JfrGTestAdaptiveSampling.bursty_rate_10_percent_vm (27 ms)
      [ RUN ] JfrGTestAdaptiveSampling.bursty_rate_90_percent_vm
      === Adaptive sampling: bursty 90%
      [ OK ] JfrGTestAdaptiveSampling.bursty_rate_90_percent_vm (27 ms)
      [----------] 8 tests from JfrGTestAdaptiveSampling (721 ms total)


      The EXPECT_NEAR expressions are hard to get right, it needs to give enough leeway to let it pass in the CI, but still test something useful in regards to the probabilistic sampling. Problem-list the test and harden.

            Unassigned Unassigned
            mgronlun Markus Grönlund
            Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

              Created:
              Updated: