Examples with VersionedSchemaId com.linkedin.databus2.schemas.VersionedSchemaId used on opensource projects

Example 1 with VersionedSchemaId

use of com.linkedin.databus2.schemas.VersionedSchemaId in project databus by linkedin.

the class TestGoldenGateEventProducer method testGGParserStats.

/**
   * test collection of parser stats, especially lag between parsed and added files
   * @throws Exception
   */
@Test
public void testGGParserStats() throws Exception {
    short[] sourceIds = new short[] { 505, 506 };
    String[] sourceNames = new String[] { "source1", "source2" };
    // setup trail Files directory
    File ggTrailDir = FileUtils.createTempDir("testGGParserStats");
    // configure physical source
    String uri = "gg://" + ggTrailDir.getAbsolutePath() + ":x3";
    PhysicalSourceStaticConfig pssc = buildSimplePssc(sourceIds, sourceNames, uri);
    LOG.info("Uri=" + uri);
    // create schema
    Schema s = Schema.parse(sourceAvroSchema);
    VersionedSchema vs = new VersionedSchema(new VersionedSchemaId("source1", (short) 3), s, null);
    // mock for schema registry
    SchemaRegistryService srs = EasyMock.createMock(SchemaRegistryService.class);
    EasyMock.expect(srs.fetchLatestVersionedSchemaBySourceName("source1")).andReturn(vs).anyTimes();
    EasyMock.expect(srs.fetchLatestVersionedSchemaBySourceName("source2")).andReturn(vs).anyTimes();
    EasyMock.expect(srs.fetchLatestVersionedSchemaBySourceName(null)).andReturn(vs);
    // mock for MaxSCNReadWriter
    MaxSCNReaderWriter mscn = EasyMock.createMock(MaxSCNReaderWriter.class);
    EasyMock.expect(mscn.getMaxScn()).andReturn((long) -2).atLeastOnce();
    mscn.saveMaxScn(EasyMock.anyLong());
    EasyMock.expectLastCall().anyTimes();
    EasyMock.replay(mscn);
    EasyMock.replay(srs);
    int totalTransWritten = 0;
    int totalFilesWritten = 0;
    // buffer
    DbusEventBufferAppendable mb = createBufMult(pssc);
    // start GG producer
    GoldenGateEventProducer gg = new GoldenGateEventProducer(pssc, srs, mb, null, mscn);
    // create first 2 files
    addToTrailFile(new File(ggTrailDir.getAbsolutePath() + "/x301"), 100, 4);
    addToTrailFile(new File(ggTrailDir.getAbsolutePath() + "/x302"), 200, 4);
    totalTransWritten = 8;
    totalFilesWritten = 2;
    // get hold of parser stats object
    final GGParserStatistics ggParserStats = gg.getParserStats();
    // all should be 0
    Assert.assertEquals(0, ggParserStats.getNumFilesParsed());
    Assert.assertEquals(0, ggParserStats.getNumFilesAdded());
    Assert.assertEquals(0, ggParserStats.getFilesLag());
    Assert.assertEquals(0, ggParserStats.getTimeLag());
    Assert.assertEquals(0, ggParserStats.getBytesLag());
    try {
        LOG.info("starting event producer");
        // -2 here does nothing. actual setting happens thru the mock of
        gg.start(-2);
        // MaxSCNReadWriter
        // let it parse first files
        TestUtil.assertWithBackoff(new ConditionCheck() {

            @Override
            public boolean check() {
                return ggParserStats.getNumFilesParsed() == 2 && (8 * _transactionPatternSize == ggParserStats.getNumBytesTotalParsed());
            }
        }, "First two files parsed", 2000, LOG);
        // stats in the interim
        Assert.assertEquals(2, ggParserStats.getNumFilesParsed());
        Assert.assertEquals(2, ggParserStats.getNumFilesAdded());
        Assert.assertEquals(0, ggParserStats.getFilesLag());
        Assert.assertEquals(0, ggParserStats.getTimeLag());
        Assert.assertEquals(0, ggParserStats.getBytesLag());
        Assert.assertEquals(totalTransWritten * _transactionPatternSize, ggParserStats.getNumBytesTotalParsed());
        gg.pause();
        // the file will get parsed but not processed
        addToTrailFile(new File(ggTrailDir.getAbsolutePath() + "/x303"), 300, 4);
        totalTransWritten += 4;
        totalFilesWritten++;
        // to get more then a ms lag time
        TestUtil.sleep(2000);
        addToTrailFile(new File(ggTrailDir.getAbsolutePath() + "/x304"), 400, 4);
        totalTransWritten += 4;
        totalFilesWritten++;
        // to guarantee we picked up stats update (stats are updated
        TestUtil.sleep(6000);
        // every 5 seconds)
        // now we should be 2 files behind. parser thread gets paused AFTER it start
        // processing the file
        // so the actuall value will be 1 file behind
        // 303(already started being parsed), only 304 is behind
        int lagFiles = 1;
        // 1 file, 4 transactions each
        long lagBytes = 1 * 4 * _transactionPatternSize;
        /*
       * Assert.assertEquals(totalFilesWritten-1, ggParserStats.getNumFilesParsed());
       * Assert.assertEquals(totalFilesWritten, ggParserStats.getNumFilesAdded());
       * Assert.assertEquals(lagFiles, ggParserStats.getFilesLag()); // because 303 got
       * parsed
       *
       * // we added 4 files and parsed 3 , so the diff should be 1 file size (4
       * trasactions in 1 file) Assert.assertEquals(lagBytes,
       * ggParserStats.getBytesLag()); Assert.assertTrue(ggParserStats.getTimeLag()>0);
       */
        gg.unpause();
        TestUtil.sleep(5000);
        // now we should catchup
        Assert.assertEquals(4, ggParserStats.getNumFilesParsed());
        Assert.assertEquals(4, ggParserStats.getNumFilesAdded());
        Assert.assertEquals(0, ggParserStats.getFilesLag());
        Assert.assertEquals(0, ggParserStats.getTimeLag());
        Assert.assertEquals(0, ggParserStats.getBytesLag());
        // append to a file
        LOG.info("pausing again");
        gg.pause();
        addToTrailFile(new File(ggTrailDir.getAbsolutePath() + "/x304"), 410, 4);
        totalTransWritten += 4;
        TestUtil.sleep(1000);
        addToTrailFile(new File(ggTrailDir.getAbsolutePath() + "/x304"), 420, 4);
        totalTransWritten += 4;
        TestUtil.sleep(2000);
        gg.unpause();
        TestUtil.sleep(5500);
        // should be still up
        Assert.assertEquals(4, ggParserStats.getNumFilesParsed());
        Assert.assertEquals(4, ggParserStats.getNumFilesAdded());
        Assert.assertEquals(0, ggParserStats.getFilesLag());
        Assert.assertEquals(0, ggParserStats.getTimeLag());
        Assert.assertEquals(0, ggParserStats.getBytesLag());
        // assert the stats
        int totalFilesSize = totalTransWritten * _transactionPatternSize;
        Assert.assertEquals((totalFilesSize / totalFilesWritten), ggParserStats.getAvgFileSize());
        Assert.assertEquals(true, ggParserStats.getAvgParseTransactionTimeNs() > 0);
        Assert.assertEquals("part1", ggParserStats.getPhysicalSourceName());
        Assert.assertEquals(totalFilesSize / totalTransWritten, ggParserStats.getAvgTransactionSize());
        Assert.assertEquals(423, ggParserStats.getMaxScn());
        // 2
        Assert.assertEquals(totalTransWritten * 2, ggParserStats.getNumTotalEvents());
        // events
        // per
        // transaction
        Assert.assertEquals(totalTransWritten, ggParserStats.getNumTransactionsTotal());
        Assert.assertEquals(totalTransWritten, ggParserStats.getNumTransactionsWithEvents());
        Assert.assertEquals(0, ggParserStats.getNumTransactionsWithoutEvents());
        Assert.assertEquals(true, ggParserStats.getTimeSinceLastAccessMs() > 0);
        Assert.assertEquals(totalTransWritten * _transactionPatternSize, ggParserStats.getNumBytesTotalParsed());
        Assert.assertEquals("NumSCNRegressions", 0, ggParserStats.getNumSCNRegressions());
        Assert.assertEquals("LastSCNRegressed", -1, ggParserStats.getLastRegressedScn());
    } finally {
        gg.shutdown();
    }
    return;
}