Github user zentol commented on a diff in the pull request: https://github.com/apache/flink/pull/2484#discussion_r80238252 --- Diff: flink-streaming-connectors/flink-connector-cassandra/src/test/java/org/apache/flink/streaming/connectors/cassandra/CassandraConnectorITCase.java --- @@ -178,89 +166,122 @@ public static void startCassandra() throws IOException { } scanner.close(); - - // Tell cassandra where the configuration files are. - // Use the test configuration file. - System.setProperty("cassandra.config", tmp.getAbsoluteFile().toURI().toString()); - if (EMBEDDED) { - cassandra = new EmbeddedCassandraService(); - cassandra.start(); + String javaCommand = getJavaCommandPath(); + + // create a logging file for the process + File tempLogFile = File.createTempFile("testlogconfig", "properties"); + CommonTestUtils.printLog4jDebugConfig(tempLogFile); + + // start the task manager process + String[] command = new String[]{ + javaCommand, + "-Dlog.level=DEBUG", + "-Dlog4j.configuration=file:" + tempLogFile.getAbsolutePath(), + "-Dcassandra.config=" + tmp.toURI(), + // these options were taken directly from the jvm.options file in the cassandra repo + "-XX:+UseThreadPriorities", + "-Xss256k", + "-XX:+AlwaysPreTouch", + "-XX:+UseTLAB", + "-XX:+ResizeTLAB", + "-XX:+UseNUMA", + "-XX:+PerfDisableSharedMem", + "-XX:+UseParNewGC", + "-XX:+UseConcMarkSweepGC", + "-XX:+CMSParallelRemarkEnabled", + "-XX:SurvivorRatio=8", + "-XX:MaxTenuringThreshold=1", + "-XX:CMSInitiatingOccupancyFraction=75", + "-XX:+UseCMSInitiatingOccupancyOnly", + "-XX:CMSWaitDuration=10000", + "-XX:+CMSParallelInitialMarkEnabled", + "-XX:+CMSEdenChunksRecordAlways", + "-XX:+CMSClassUnloadingEnabled", + + "-classpath", getCurrentClasspath(), + EmbeddedCassandraService.class.getName() + }; + + ProcessBuilder bld = new ProcessBuilder(command); + cassandra = bld.start(); + sw = new StringWriter(); + new PipeForwarder(cassandra.getErrorStream(), sw); } - try { - Thread.sleep(1000 * 10); - } catch (InterruptedException e) { //give cassandra a few seconds to start up + int attempt = 0; + while (true) { + try { + attempt++; + cluster = builder.getCluster(); + session = cluster.connect(); + break; + } catch (Exception e) { + if (attempt > 30) { + throw e; + } + try { + Thread.sleep(1000); + } catch (InterruptedException e1) { + } + } } - - cluster = builder.getCluster(); - session = cluster.connect(); + LOG.debug("Connection established after " + attempt + " attempts."); session.execute(CREATE_KEYSPACE_QUERY); - session.execute(CREATE_TABLE_QUERY); - } - - @BeforeClass - public static void startFlink() throws Exception { - Configuration config = new Configuration(); - config.setInteger(ConfigConstants.TASK_MANAGER_NUM_TASK_SLOTS, 4); + session.execute(CREATE_TABLE_QUERY.replace("$TABLE", "flink_initial")); - flinkCluster = new ForkableFlinkMiniCluster(config); - flinkCluster.start(); + try { + Thread.sleep(5000); --- End diff -- I tried that and the results were actually quite interesting. The last issues i dealt with was that fairly regularly the first test that is executed would fail. I could disable one test after another, but the failure would just occur on the test that was now executed first. I figured that Cassandra may not be ready yet to receive data, so i implemented what you suggested: create a table, write a row, and check whether it was written. That thing failed not even once, but sure enough the tests still did.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---