return
    }

    try {
      testHttpUrl(testData)
      if (KNOWN_FAILURES.contains(testData.toString())) {
        System.err.println("Expected failure but was success: $testData")
      }
    } catch (e: Throwable) {
      if (KNOWN_FAILURES.contains(testData.toString())) {
        System.err.println("Ignoring known failure: $testData")
        e.printStackTrace()
      } else {
        throw e
      }
    }

In 12 drive example above, with MinIO server running in the default configuration, you can lose any of the six drives and still reconstruct the data reliably from the remaining drives.

## Why is Erasure Code useful?

      description: |
        Please only use this template for submitting reports about flaky tests or jobs (pass or fail with no underlying change in code) in Kubernetes CI.
        Links to go.k8s.io/triage and/or links to specific failures in spyglass are appreciated.
        Please see the deflaking doc (https://github.com/kubernetes/community/blob/master/contributors/devel/sig-testing/flaky-tests.md) for more guidance.
    validations:
      required: true

  // `validation.policy.admission.k8s.io/validation_failure` audit annotation
  // with a value containing the details of the validation failures, formatted as
  // a JSON list of objects, each with the following fields:
  // - message: The validation failure message string
  // - policy: The resource name of the ValidatingAdmissionPolicy
  // - binding: The resource name of the ValidatingAdmissionPolicyBinding

  awk '/Streaming build results to/ {print $NF}' "$TFCI_OUTPUT_DIR/script.log" | uniq
}

# Print out any ResultStore URLs for Bazel invocations' results.
# Each failed target there will have its own representation, making failures
# easier to find and read.
function resultstore_extract {
  local \
    XML_PATH="$TFCI_OUTPUT_DIR/Bazel_Test_and_Build_Results/sponge_log.xml"

  python3 \

    e is SSLHandshakeException && e.cause is CertificateException -> false

    // e.g. a certificate pinning error.
    e is SSLPeerUnverifiedException -> false

    // Retry for all other SSL failures.
    e is SSLException -> true

    else -> false
  }

   * simulate late connection pool failures.
   */
  object DisconnectAfterRequest : SocketPolicy

  /**
   * Half close connection (InputStream for client) after reading the request but before
   * writing the response. Use this to simulate late connection pool failures.
   */
  object HalfCloseAfterRequest : SocketPolicy

# See the License for the specific language governing permissions and
# limitations under the License.

# retry.sh retries a command until it succeeds. It accepts a regex pattern to match failures on to
# determine if a retry should be attempted.
# Example: retry.sh "connection timed out" ./my-flaky-script.sh some args
# This will run "my-flaky-script.sh", retrying any failed runs that output "connection timed out" up

        .isTrue();
    if (!failures.isEmpty()) {
      StringWriter message = new StringWriter();
      PrintWriter writer = new PrintWriter(message);
      writer.println("Expected no failures, but found:");
      for (AssertionError failure : failures) {
        failure.printStackTrace(writer);
      }
      failures.clear();
      assertWithMessage(message.toString()).fail();
    }

		case tierRequestsFailure:
			fail += metric.Value
		}
	}
	if int(succ) != expSuccess {
		t.Fatalf("Expected %d successes but got %f", expSuccess, succ)
	}
	if int(fail) != expFailure {
		t.Fatalf("Expected %d failures but got %f", expFailure, fail)
	}