Search in sources :

Example 46 with RecordHeaders

use of org.apache.kafka.common.header.internals.RecordHeaders in project kafka by apache.

the class TimeOrderedKeyValueBufferTest method shouldRestoreV3Format.

@Test
public void shouldRestoreV3Format() {
    final TimeOrderedKeyValueBuffer<String, String> buffer = bufferSupplier.apply(testName);
    final MockInternalProcessorContext context = makeContext();
    buffer.init((StateStoreContext) context, buffer);
    final RecordBatchingStateRestoreCallback stateRestoreCallback = (RecordBatchingStateRestoreCallback) context.stateRestoreCallback(testName);
    context.setRecordContext(new ProcessorRecordContext(0, 0, 0, "", new RecordHeaders()));
    final RecordHeaders headers = new RecordHeaders(new Header[] { new RecordHeader("v", new byte[] { (byte) 3 }) });
    // These serialized formats were captured by running version 2.4 code.
    // They verify that an upgrade from 2.4 will work.
    // Do not change them.
    final String toDeleteBinary = "0000000000000000000000000000000000000005746F70696300000000FFFFFFFFFFFFFFFFFFFFFFFF00000006646F6F6D65640000000000000000";
    final String asdfBinary = "0000000000000001000000000000000000000005746F70696300000000FFFFFFFFFFFFFFFFFFFFFFFF00000004717765720000000000000002";
    final String zxcvBinary1 = "0000000000000002000000000000000000000005746F70696300000000FFFFFFFF0000000870726576696F75730000000749474E4F52454400000005336F34696D0000000000000001";
    final String zxcvBinary2 = "0000000000000003000000000000000000000005746F70696300000000FFFFFFFF0000000870726576696F757300000005336F34696D000000046E6578740000000000000001";
    stateRestoreCallback.restoreBatch(asList(new ConsumerRecord<>("changelog-topic", 0, 0, 999, TimestampType.CREATE_TIME, -1, -1, "todelete".getBytes(UTF_8), hexStringToByteArray(toDeleteBinary), headers, Optional.empty()), new ConsumerRecord<>("changelog-topic", 0, 1, 9999, TimestampType.CREATE_TIME, -1, -1, "asdf".getBytes(UTF_8), hexStringToByteArray(asdfBinary), headers, Optional.empty()), new ConsumerRecord<>("changelog-topic", 0, 2, 99, TimestampType.CREATE_TIME, -1, -1, "zxcv".getBytes(UTF_8), hexStringToByteArray(zxcvBinary1), headers, Optional.empty()), new ConsumerRecord<>("changelog-topic", 0, 2, 100, TimestampType.CREATE_TIME, -1, -1, "zxcv".getBytes(UTF_8), hexStringToByteArray(zxcvBinary2), headers, Optional.empty())));
    assertThat(buffer.numRecords(), is(3));
    assertThat(buffer.minTimestamp(), is(0L));
    assertThat(buffer.bufferSize(), is(142L));
    stateRestoreCallback.restoreBatch(singletonList(new ConsumerRecord<>("changelog-topic", 0, 3, 3, TimestampType.CREATE_TIME, -1, -1, "todelete".getBytes(UTF_8), null, new RecordHeaders(), Optional.empty())));
    assertThat(buffer.numRecords(), is(2));
    assertThat(buffer.minTimestamp(), is(1L));
    assertThat(buffer.bufferSize(), is(95L));
    assertThat(buffer.priorValueForBuffered("todelete"), is(Maybe.undefined()));
    assertThat(buffer.priorValueForBuffered("asdf"), is(Maybe.defined(null)));
    assertThat(buffer.priorValueForBuffered("zxcv"), is(Maybe.defined(ValueAndTimestamp.make("previous", -1))));
    // flush the buffer into a list in buffer order so we can make assertions about the contents.
    final List<Eviction<String, String>> evicted = new LinkedList<>();
    buffer.evictWhile(() -> true, evicted::add);
    // Several things to note:
    // * The buffered records are ordered according to their buffer time (serialized in the value of the changelog)
    // * The record timestamps are properly restored, and not conflated with the record's buffer time.
    // * The keys and values are properly restored
    // * The record topic is set to the original input topic, *not* the changelog topic
    // * The record offset preserves the original input record's offset, *not* the offset of the changelog record
    assertThat(evicted, is(asList(new Eviction<>("zxcv", new Change<>("next", "3o4im"), getContext(3L)), new Eviction<>("asdf", new Change<>("qwer", null), getContext(1L)))));
    cleanup(context, buffer);
}
Also used : Change(org.apache.kafka.streams.kstream.internals.Change) ConsumerRecord(org.apache.kafka.clients.consumer.ConsumerRecord) LinkedList(java.util.LinkedList) RecordBatchingStateRestoreCallback(org.apache.kafka.streams.processor.internals.RecordBatchingStateRestoreCallback) RecordHeaders(org.apache.kafka.common.header.internals.RecordHeaders) ProcessorRecordContext(org.apache.kafka.streams.processor.internals.ProcessorRecordContext) Eviction(org.apache.kafka.streams.state.internals.TimeOrderedKeyValueBuffer.Eviction) MockInternalProcessorContext(org.apache.kafka.test.MockInternalProcessorContext) RecordHeader(org.apache.kafka.common.header.internals.RecordHeader) Test(org.junit.Test)

Example 47 with RecordHeaders

use of org.apache.kafka.common.header.internals.RecordHeaders in project kafka by apache.

the class TimeOrderedKeyValueBufferTest method shouldFlush.

@Test
public void shouldFlush() {
    final TimeOrderedKeyValueBuffer<String, String> buffer = bufferSupplier.apply(testName);
    final MockInternalProcessorContext context = makeContext();
    buffer.init((StateStoreContext) context, buffer);
    putRecord(buffer, context, 2L, 0L, "asdf", "2093j");
    putRecord(buffer, context, 1L, 1L, "zxcv", "3gon4i");
    putRecord(buffer, context, 0L, 2L, "deleteme", "deadbeef");
    // replace "deleteme" with a tombstone
    buffer.evictWhile(() -> buffer.minTimestamp() < 1, kv -> {
    });
    // flush everything to the changelog
    buffer.flush();
    // the buffer should serialize the buffer time and the value as byte[],
    // which we can't compare for equality using ProducerRecord.
    // As a workaround, I'm deserializing them and shoving them in a KeyValue, just for ease of testing.
    final List<ProducerRecord<String, KeyValue<Long, BufferValue>>> collected = ((MockRecordCollector) context.recordCollector()).collected().stream().map(pr -> {
        final KeyValue<Long, BufferValue> niceValue;
        if (pr.value() == null) {
            niceValue = null;
        } else {
            final byte[] serializedValue = (byte[]) pr.value();
            final ByteBuffer valueBuffer = ByteBuffer.wrap(serializedValue);
            final BufferValue contextualRecord = BufferValue.deserialize(valueBuffer);
            final long timestamp = valueBuffer.getLong();
            niceValue = new KeyValue<>(timestamp, contextualRecord);
        }
        return new ProducerRecord<>(pr.topic(), pr.partition(), pr.timestamp(), pr.key().toString(), niceValue, pr.headers());
    }).collect(Collectors.toList());
    assertThat(collected, is(asList(new ProducerRecord<>(APP_ID + "-" + testName + "-changelog", // Producer will assign
    0, null, "deleteme", null, new RecordHeaders()), new ProducerRecord<>(APP_ID + "-" + testName + "-changelog", 0, null, "zxcv", new KeyValue<>(1L, getBufferValue("3gon4i", 1)), CHANGELOG_HEADERS), new ProducerRecord<>(APP_ID + "-" + testName + "-changelog", 0, null, "asdf", new KeyValue<>(2L, getBufferValue("2093j", 0)), CHANGELOG_HEADERS))));
    cleanup(context, buffer);
}
Also used : StreamsConfig(org.apache.kafka.streams.StreamsConfig) ProducerRecord(org.apache.kafka.clients.producer.ProducerRecord) TaskId(org.apache.kafka.streams.processor.TaskId) ProcessorRecordContext(org.apache.kafka.streams.processor.internals.ProcessorRecordContext) RunWith(org.junit.runner.RunWith) Random(java.util.Random) Eviction(org.apache.kafka.streams.state.internals.TimeOrderedKeyValueBuffer.Eviction) RecordHeader(org.apache.kafka.common.header.internals.RecordHeader) ValueAndTimestamp(org.apache.kafka.streams.state.ValueAndTimestamp) Function(java.util.function.Function) RecordBatchingStateRestoreCallback(org.apache.kafka.streams.processor.internals.RecordBatchingStateRestoreCallback) ByteBuffer(java.nio.ByteBuffer) Collections.singletonList(java.util.Collections.singletonList) RecordHeaders(org.apache.kafka.common.header.internals.RecordHeaders) StringDeserializer(org.apache.kafka.common.serialization.StringDeserializer) StateStoreContext(org.apache.kafka.streams.processor.StateStoreContext) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) Arrays.asList(java.util.Arrays.asList) Serdes(org.apache.kafka.common.serialization.Serdes) StringSerializer(org.apache.kafka.common.serialization.StringSerializer) Record(org.apache.kafka.streams.processor.api.Record) Assert.fail(org.junit.Assert.fail) MockRecordCollector(org.apache.kafka.test.MockRecordCollector) MatcherAssert.assertThat(org.hamcrest.MatcherAssert.assertThat) LinkedList(java.util.LinkedList) TimestampType(org.apache.kafka.common.record.TimestampType) Parameterized(org.junit.runners.Parameterized) Utils(org.apache.kafka.common.utils.Utils) Properties(java.util.Properties) TestUtils(org.apache.kafka.test.TestUtils) UTF_8(java.nio.charset.StandardCharsets.UTF_8) Collection(java.util.Collection) KeyValue(org.apache.kafka.streams.KeyValue) CHANGELOG_HEADERS(org.apache.kafka.streams.state.internals.InMemoryTimeOrderedKeyValueBuffer.CHANGELOG_HEADERS) Test(org.junit.Test) IOException(java.io.IOException) Collectors(java.util.stream.Collectors) MockInternalProcessorContext(org.apache.kafka.test.MockInternalProcessorContext) List(java.util.List) Header(org.apache.kafka.common.header.Header) ConsumerRecord(org.apache.kafka.clients.consumer.ConsumerRecord) Optional(java.util.Optional) Matchers.is(org.hamcrest.Matchers.is) Change(org.apache.kafka.streams.kstream.internals.Change) KeyValue(org.apache.kafka.streams.KeyValue) RecordHeaders(org.apache.kafka.common.header.internals.RecordHeaders) ProducerRecord(org.apache.kafka.clients.producer.ProducerRecord) MockInternalProcessorContext(org.apache.kafka.test.MockInternalProcessorContext) ByteBuffer(java.nio.ByteBuffer) Test(org.junit.Test)

Example 48 with RecordHeaders

use of org.apache.kafka.common.header.internals.RecordHeaders in project kafka by apache.

the class TimeOrderedKeyValueBufferTest method shouldRestoreV1Format.

@Test
public void shouldRestoreV1Format() {
    final TimeOrderedKeyValueBuffer<String, String> buffer = bufferSupplier.apply(testName);
    final MockInternalProcessorContext context = makeContext();
    buffer.init((StateStoreContext) context, buffer);
    final RecordBatchingStateRestoreCallback stateRestoreCallback = (RecordBatchingStateRestoreCallback) context.stateRestoreCallback(testName);
    context.setRecordContext(new ProcessorRecordContext(0, 0, 0, "", new RecordHeaders()));
    final RecordHeaders v1FlagHeaders = new RecordHeaders(new Header[] { new RecordHeader("v", new byte[] { (byte) 1 }) });
    // These serialized formats were captured by running version 2.2 code.
    // They verify that an upgrade from 2.2 will work.
    // Do not change them.
    final String toDeleteBinary = "00000000000000000000000000000000000000000000000000000005746F70696300000000FFFFFFFF0000000EFFFFFFFF00000006646F6F6D6564";
    final String asdfBinary = "00000000000000020000000000000001000000000000000000000005746F70696300000000FFFFFFFF0000000CFFFFFFFF0000000471776572";
    final String zxcvBinary1 = "00000000000000010000000000000002000000000000000000000005746F70696300000000FFFFFFFF000000150000000870726576696F757300000005336F34696D";
    final String zxcvBinary2 = "00000000000000010000000000000003000000000000000000000005746F70696300000000FFFFFFFF0000001100000005336F34696D000000046E657874";
    stateRestoreCallback.restoreBatch(asList(new ConsumerRecord<>("changelog-topic", 0, 0, 999, TimestampType.CREATE_TIME, -1, -1, "todelete".getBytes(UTF_8), hexStringToByteArray(toDeleteBinary), v1FlagHeaders, Optional.empty()), new ConsumerRecord<>("changelog-topic", 0, 1, 9999, TimestampType.CREATE_TIME, -1, -1, "asdf".getBytes(UTF_8), hexStringToByteArray(asdfBinary), v1FlagHeaders, Optional.empty()), new ConsumerRecord<>("changelog-topic", 0, 2, 99, TimestampType.CREATE_TIME, -1, -1, "zxcv".getBytes(UTF_8), hexStringToByteArray(zxcvBinary1), v1FlagHeaders, Optional.empty()), new ConsumerRecord<>("changelog-topic", 0, 3, 100, TimestampType.CREATE_TIME, -1, -1, "zxcv".getBytes(UTF_8), hexStringToByteArray(zxcvBinary2), v1FlagHeaders, Optional.empty())));
    assertThat(buffer.numRecords(), is(3));
    assertThat(buffer.minTimestamp(), is(0L));
    assertThat(buffer.bufferSize(), is(142L));
    stateRestoreCallback.restoreBatch(singletonList(new ConsumerRecord<>("changelog-topic", 0, 3, 3, TimestampType.CREATE_TIME, -1, -1, "todelete".getBytes(UTF_8), null, new RecordHeaders(), Optional.empty())));
    assertThat(buffer.numRecords(), is(2));
    assertThat(buffer.minTimestamp(), is(1L));
    assertThat(buffer.bufferSize(), is(95L));
    assertThat(buffer.priorValueForBuffered("todelete"), is(Maybe.undefined()));
    assertThat(buffer.priorValueForBuffered("asdf"), is(Maybe.defined(null)));
    assertThat(buffer.priorValueForBuffered("zxcv"), is(Maybe.defined(ValueAndTimestamp.make("previous", -1))));
    // flush the buffer into a list in buffer order so we can make assertions about the contents.
    final List<Eviction<String, String>> evicted = new LinkedList<>();
    buffer.evictWhile(() -> true, evicted::add);
    // Several things to note:
    // * The buffered records are ordered according to their buffer time (serialized in the value of the changelog)
    // * The record timestamps are properly restored, and not conflated with the record's buffer time.
    // * The keys and values are properly restored
    // * The record topic is set to the original input topic, *not* the changelog topic
    // * The record offset preserves the original input record's offset, *not* the offset of the changelog record
    assertThat(evicted, is(asList(new Eviction<>("zxcv", new Change<>("next", "3o4im"), getContext(3L)), new Eviction<>("asdf", new Change<>("qwer", null), getContext(1L)))));
    cleanup(context, buffer);
}
Also used : Change(org.apache.kafka.streams.kstream.internals.Change) ConsumerRecord(org.apache.kafka.clients.consumer.ConsumerRecord) LinkedList(java.util.LinkedList) RecordBatchingStateRestoreCallback(org.apache.kafka.streams.processor.internals.RecordBatchingStateRestoreCallback) RecordHeaders(org.apache.kafka.common.header.internals.RecordHeaders) ProcessorRecordContext(org.apache.kafka.streams.processor.internals.ProcessorRecordContext) Eviction(org.apache.kafka.streams.state.internals.TimeOrderedKeyValueBuffer.Eviction) MockInternalProcessorContext(org.apache.kafka.test.MockInternalProcessorContext) RecordHeader(org.apache.kafka.common.header.internals.RecordHeader) Test(org.junit.Test)

Example 49 with RecordHeaders

use of org.apache.kafka.common.header.internals.RecordHeaders in project kafka by apache.

the class TimeOrderedKeyValueBufferTest method shouldRestoreOldUnversionedFormat.

@Test
public void shouldRestoreOldUnversionedFormat() {
    final TimeOrderedKeyValueBuffer<String, String> buffer = bufferSupplier.apply(testName);
    final MockInternalProcessorContext context = makeContext();
    buffer.init((StateStoreContext) context, buffer);
    final RecordBatchingStateRestoreCallback stateRestoreCallback = (RecordBatchingStateRestoreCallback) context.stateRestoreCallback(testName);
    context.setRecordContext(new ProcessorRecordContext(0, 0, 0, "", new RecordHeaders()));
    // These serialized formats were captured by running version 2.1 code.
    // They verify that an upgrade from 2.1 will work.
    // Do not change them.
    final String toDeleteBinaryValue = "0000000000000000FFFFFFFF00000006646F6F6D6564";
    final String asdfBinaryValue = "0000000000000002FFFFFFFF0000000471776572";
    final String zxcvBinaryValue1 = "00000000000000010000000870726576696F757300000005656F34696D";
    final String zxcvBinaryValue2 = "000000000000000100000005656F34696D000000046E657874";
    stateRestoreCallback.restoreBatch(asList(new ConsumerRecord<>("changelog-topic", 0, 0, 0, TimestampType.CREATE_TIME, -1, -1, "todelete".getBytes(UTF_8), hexStringToByteArray(toDeleteBinaryValue), new RecordHeaders(), Optional.empty()), new ConsumerRecord<>("changelog-topic", 0, 1, 1, TimestampType.CREATE_TIME, -1, -1, "asdf".getBytes(UTF_8), hexStringToByteArray(asdfBinaryValue), new RecordHeaders(), Optional.empty()), new ConsumerRecord<>("changelog-topic", 0, 2, 2, TimestampType.CREATE_TIME, -1, -1, "zxcv".getBytes(UTF_8), hexStringToByteArray(zxcvBinaryValue1), new RecordHeaders(), Optional.empty()), new ConsumerRecord<>("changelog-topic", 0, 3, 3, TimestampType.CREATE_TIME, -1, -1, "zxcv".getBytes(UTF_8), hexStringToByteArray(zxcvBinaryValue2), new RecordHeaders(), Optional.empty())));
    assertThat(buffer.numRecords(), is(3));
    assertThat(buffer.minTimestamp(), is(0L));
    assertThat(buffer.bufferSize(), is(172L));
    stateRestoreCallback.restoreBatch(singletonList(new ConsumerRecord<>("changelog-topic", 0, 3, 3, TimestampType.CREATE_TIME, -1, -1, "todelete".getBytes(UTF_8), null, new RecordHeaders(), Optional.empty())));
    assertThat(buffer.numRecords(), is(2));
    assertThat(buffer.minTimestamp(), is(1L));
    assertThat(buffer.bufferSize(), is(115L));
    assertThat(buffer.priorValueForBuffered("todelete"), is(Maybe.undefined()));
    assertThat(buffer.priorValueForBuffered("asdf"), is(Maybe.defined(null)));
    assertThat(buffer.priorValueForBuffered("zxcv"), is(Maybe.defined(ValueAndTimestamp.make("previous", -1))));
    // flush the buffer into a list in buffer order so we can make assertions about the contents.
    final List<Eviction<String, String>> evicted = new LinkedList<>();
    buffer.evictWhile(() -> true, evicted::add);
    // Several things to note:
    // * The buffered records are ordered according to their buffer time (serialized in the value of the changelog)
    // * The record timestamps are properly restored, and not conflated with the record's buffer time.
    // * The keys and values are properly restored
    // * The record topic is set to the changelog topic. This was an oversight in the original implementation,
    // which is fixed in changelog format v1. But upgraded applications still need to be able to handle the
    // original format.
    assertThat(evicted, is(asList(new Eviction<>("zxcv", new Change<>("next", "eo4im"), new ProcessorRecordContext(3L, 3, 0, "changelog-topic", new RecordHeaders())), new Eviction<>("asdf", new Change<>("qwer", null), new ProcessorRecordContext(1L, 1, 0, "changelog-topic", new RecordHeaders())))));
    cleanup(context, buffer);
}
Also used : Change(org.apache.kafka.streams.kstream.internals.Change) ConsumerRecord(org.apache.kafka.clients.consumer.ConsumerRecord) LinkedList(java.util.LinkedList) RecordBatchingStateRestoreCallback(org.apache.kafka.streams.processor.internals.RecordBatchingStateRestoreCallback) RecordHeaders(org.apache.kafka.common.header.internals.RecordHeaders) ProcessorRecordContext(org.apache.kafka.streams.processor.internals.ProcessorRecordContext) Eviction(org.apache.kafka.streams.state.internals.TimeOrderedKeyValueBuffer.Eviction) MockInternalProcessorContext(org.apache.kafka.test.MockInternalProcessorContext) Test(org.junit.Test)

Example 50 with RecordHeaders

use of org.apache.kafka.common.header.internals.RecordHeaders in project kafka by apache.

the class TimeOrderedKeyValueBufferTest method shouldNotRestoreUnrecognizedVersionRecord.

@Test
public void shouldNotRestoreUnrecognizedVersionRecord() {
    final TimeOrderedKeyValueBuffer<String, String> buffer = bufferSupplier.apply(testName);
    final MockInternalProcessorContext context = makeContext();
    buffer.init((StateStoreContext) context, buffer);
    final RecordBatchingStateRestoreCallback stateRestoreCallback = (RecordBatchingStateRestoreCallback) context.stateRestoreCallback(testName);
    context.setRecordContext(new ProcessorRecordContext(0, 0, 0, "", new RecordHeaders()));
    final RecordHeaders unknownFlagHeaders = new RecordHeaders(new Header[] { new RecordHeader("v", new byte[] { (byte) -1 }) });
    final byte[] todeleteValue = getBufferValue("doomed", 0).serialize(0).array();
    try {
        stateRestoreCallback.restoreBatch(singletonList(new ConsumerRecord<>("changelog-topic", 0, 0, 999, TimestampType.CREATE_TIME, -1, -1, "todelete".getBytes(UTF_8), ByteBuffer.allocate(Long.BYTES + todeleteValue.length).putLong(0L).put(todeleteValue).array(), unknownFlagHeaders, Optional.empty())));
        fail("expected an exception");
    } catch (final IllegalArgumentException expected) {
    // nothing to do.
    } finally {
        cleanup(context, buffer);
    }
}
Also used : RecordBatchingStateRestoreCallback(org.apache.kafka.streams.processor.internals.RecordBatchingStateRestoreCallback) RecordHeaders(org.apache.kafka.common.header.internals.RecordHeaders) ProcessorRecordContext(org.apache.kafka.streams.processor.internals.ProcessorRecordContext) MockInternalProcessorContext(org.apache.kafka.test.MockInternalProcessorContext) RecordHeader(org.apache.kafka.common.header.internals.RecordHeader) ConsumerRecord(org.apache.kafka.clients.consumer.ConsumerRecord) Test(org.junit.Test)

Aggregations

RecordHeaders (org.apache.kafka.common.header.internals.RecordHeaders)149 Test (org.junit.Test)107 ConsumerRecord (org.apache.kafka.clients.consumer.ConsumerRecord)49 ProcessorRecordContext (org.apache.kafka.streams.processor.internals.ProcessorRecordContext)41 Headers (org.apache.kafka.common.header.Headers)33 RecordHeader (org.apache.kafka.common.header.internals.RecordHeader)24 PrepareForTest (org.powermock.core.classloader.annotations.PrepareForTest)24 TopicPartition (org.apache.kafka.common.TopicPartition)22 Position (org.apache.kafka.streams.query.Position)17 ArrayList (java.util.ArrayList)12 HashMap (java.util.HashMap)12 ByteBuffer (java.nio.ByteBuffer)11 Struct (org.apache.kafka.connect.data.Struct)11 Test (org.junit.jupiter.api.Test)11 Header (org.apache.kafka.common.header.Header)10 LinkedHashMap (java.util.LinkedHashMap)9 Bytes (org.apache.kafka.common.utils.Bytes)9 StreamsException (org.apache.kafka.streams.errors.StreamsException)9 ProducerRecord (org.apache.kafka.clients.producer.ProducerRecord)8 Metrics (org.apache.kafka.common.metrics.Metrics)8