Rerun ReplicationAwareTokenAllocatorTest on failure to avoid flakiness.

patch by Branimir Lambov; reviewed by Sylvain Lebresne for CASSANDRA-12277.
This commit is contained in:
Branimir Lambov 2016-07-26 15:19:51 +03:00 committed by Sylvain Lebresne
parent 3f9b9ffe6b
commit a5cbb02619
3 changed files with 72 additions and 0 deletions

View File

@ -1,4 +1,5 @@
3.0.9
* Rerun ReplicationAwareTokenAllocatorTest on failure to avoid flakiness (CASSANDRA-12277)
* Exception when computing read-repair for range tombstones (CASSANDRA-12263)
* Lost counter writes in compact table and static columns (CASSANDRA-12219)
* AssertionError with MVs on updating a row that isn't indexed due to a null value (CASSANDRA-12247)

View File

@ -29,6 +29,7 @@ import org.apache.commons.math3.stat.descriptive.SummaryStatistics;
import org.junit.Test;
import org.apache.cassandra.Util;
import org.apache.cassandra.dht.Murmur3Partitioner;
import org.apache.cassandra.dht.Token;
@ -545,6 +546,20 @@ public class ReplicationAwareTokenAllocatorTest
@Test
public void testNewCluster()
{
Util.flakyTest(this::flakyTestNewCluster,
5,
"It tends to fail sometimes due to the random selection of the tokens in the first few nodes.");
}
public void flakyTestNewCluster()
{
// This test is flaky because the selection of the tokens for the first RF nodes (which is random, with an
// uncontrolled seed) can sometimes cause a pathological situation where the algorithm will find a (close to)
// ideal distribution of tokens for some number of nodes, which in turn will inevitably cause it to go into a
// bad (unacceptable to the test criteria) distribution after adding one more node.
// This should happen very rarely, unless something is broken in the token allocation code.
for (int rf = 2; rf <= 5; ++rf)
{
for (int perUnitCount = 1; perUnitCount <= MAX_VNODE_COUNT; perUnitCount *= 4)

View File

@ -536,6 +536,62 @@ public class Util
thread.join(10000);
}
public static AssertionError runCatchingAssertionError(Runnable test)
{
try
{
test.run();
return null;
}
catch (AssertionError e)
{
return e;
}
}
/**
* Wrapper function used to run a test that can sometimes flake for uncontrollable reasons.
*
* If the given test fails on the first run, it is executed the given number of times again, expecting all secondary
* runs to succeed. If they do, the failure is understood as a flake and the test is treated as passing.
*
* Do not use this if the test is deterministic and its success is not influenced by external factors (such as time,
* selection of random seed, network failures, etc.). If the test can be made independent of such factors, it is
* probably preferable to do so rather than use this method.
*
* @param test The test to run.
* @param rerunsOnFailure How many times to re-run it if it fails. All reruns must pass.
* @param message Message to send to System.err on initial failure.
*/
public static void flakyTest(Runnable test, int rerunsOnFailure, String message)
{
AssertionError e = runCatchingAssertionError(test);
if (e == null)
return; // success
System.err.format("Test failed. %s%n"
+ "Re-running %d times to verify it isn't failing more often than it should.%n"
+ "Failure was: %s%n", message, rerunsOnFailure, e);
e.printStackTrace();
int rerunsFailed = 0;
for (int i = 0; i < rerunsOnFailure; ++i)
{
AssertionError t = runCatchingAssertionError(test);
if (t != null)
{
++rerunsFailed;
e.addSuppressed(t);
}
}
if (rerunsFailed > 0)
{
System.err.format("Test failed in %d of the %d reruns.%n", rerunsFailed, rerunsOnFailure);
throw e;
}
System.err.println("All reruns succeeded. Failure treated as flake.");
}
// for use with Optional in tests, can be used as an argument to orElseThrow
public static Supplier<AssertionError> throwAssert(final String message)
{