//JAVA TO C# CONVERTER TODO TASK: Most Java annotations will not have direct .NET equivalent attributes:
//ORIGINAL LINE: @Ignore("This belongs better in a benchmarking suite.") @Test public void electionPerformance_NormalConditions() throws Throwable
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
public virtual void ElectionPerformanceNormalConditions()
{
/* This test runs with with few iterations. Hence it does not have the power to catch
* regressions efficiently. Its purpose is mainly to run elections using real-world
* parameters and catch very obvious regressions while not contributing overly much to the
* regression test suites total runtime. */
// given parameters
const long networkLatency = 15L;
const long electionTimeout = 500L;
const long heartbeatInterval = 250L;
const int iterations = 10;
TestNetwork net = new TestNetwork <>((i, o) => networkLatency);
ISet <MemberId> members = asSet(member(0), member(1), member(2));
Fixture fixture = new Fixture(members, net, electionTimeout, heartbeatInterval);
DisconnectLeaderScenario scenario = new DisconnectLeaderScenario(fixture, electionTimeout);
try
{
// when running scenario
fixture.Boot();
scenario.Run(iterations, 10 * electionTimeout);
}
finally
{
fixture.TearDown();
}
DisconnectLeaderScenario.Result result = scenario.Result();
/* These bounds have been experimentally established and should have a very low
* likelihood for false positives without an actual major regression. If this test fails
* then the recommended action is to run the test manually and interpret the results
* to guide further action. Perhaps the power of the test has to be improved, but
* the intention here is not to catch anything but the most major of regressions. */
assertThat(result.NonCollidingAverage, lessThan(2.0 * electionTimeout));
if (result.CollisionCount > 3)
{
assertThat(result.CollidingAverage, lessThan(6.0 * electionTimeout));
}
assertThat(result.TimeoutCount, @is(0L));
}
//JAVA TO C# CONVERTER TODO TASK: Most Java annotations will not have direct .NET equivalent attributes:
//ORIGINAL LINE: @Ignore("This belongs better in a benchmarking suite.") @Test public void electionPerformance_RapidConditions() throws Throwable
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in C#:
public virtual void ElectionPerformanceRapidConditions()
{
// given parameters
const long networkLatency = 1L;
const long electionTimeout = 30L;
const long heartbeatInterval = 15L;
const int iterations = 100;
TestNetwork net = new TestNetwork <>((i, o) => networkLatency);
ISet <MemberId> members = asSet(member(0), member(1), member(2));
Fixture fixture = new Fixture(members, net, electionTimeout, heartbeatInterval);
DisconnectLeaderScenario scenario = new DisconnectLeaderScenario(fixture, electionTimeout);
try
{
// when running scenario
fixture.Boot();
scenario.Run(iterations, 10 * electionTimeout);
}
finally
{
fixture.TearDown();
}
DisconnectLeaderScenario.Result result = scenario.Result();
/* These bounds have been experimentally established and should have a very low
* likelihood for false positives without an actual major regression. If this test fails
* then the recommended action is to run the test manually and interpret the results
* to guide further action. Perhaps the power of the test has to be improved, but
* the intention here is not to catch anything but the most major of regressions. */
assertThat(result.NonCollidingAverage, lessThan(2.0 * electionTimeout));
// because of the high number of iterations, it is possible to assert on the collision rate
assertThat(result.CollisionRate, lessThan(0.50d));
if (result.CollisionCount > 10)
{
assertThat(result.CollidingAverage, lessThan(5.0 * electionTimeout));
}
assertThat(result.TimeoutCount, lessThanOrEqualTo(1L)); // for GC or whatever reason
}
