Skip to content

Commit

Permalink
remove option to transform message in the indexer
Browse files Browse the repository at this point in the history
  • Loading branch information
@vthacker
vthacker committed Feb 29, 2024
1 parent 6d8760c commit 8332866
Show file tree
Hide file tree
Showing 24 changed files with 649 additions and 731 deletions.
375 changes: 188 additions & 187 deletions benchmarks/src/main/java/com/slack/kaldb/IndexAPILog.java
View file
Original file line number Diff line number Diff line change
@@ -1,187 +1,188 @@
package com.slack.kaldb;

import com.google.protobuf.ByteString;
import com.slack.kaldb.logstore.LogMessage;
import com.slack.kaldb.logstore.LuceneIndexStoreImpl;
import com.slack.kaldb.logstore.schema.SchemaAwareLogDocumentBuilderImpl;
import com.slack.kaldb.writer.LogMessageWriterImpl;
import com.slack.service.murron.Murron;
import io.micrometer.core.instrument.MeterRegistry;
import io.micrometer.core.instrument.simple.SimpleMeterRegistry;
import java.io.BufferedReader;
import java.io.File;
import java.io.IOException;
import java.nio.charset.StandardCharsets;
import java.nio.file.Files;
import java.nio.file.Path;
import java.nio.file.Paths;
import java.text.SimpleDateFormat;
import java.time.Duration;
import java.util.Comparator;
import java.util.Random;
import java.util.stream.Stream;
import org.apache.commons.io.FileUtils;
import org.apache.kafka.clients.consumer.ConsumerRecord;
import org.apache.kafka.common.record.TimestampType;
import org.apache.lucene.store.Directory;
import org.openjdk.jmh.annotations.*;

@State(Scope.Thread)
public class IndexAPILog {

private Random random;
private final Duration commitInterval = Duration.ofSeconds(5 * 60);
private final Duration refreshInterval = Duration.ofSeconds(5 * 60);

private Path tempDirectory;
private MeterRegistry registry;
LuceneIndexStoreImpl logStore;

private String apiLogFile;
private BufferedReader reader;
private static SimpleDateFormat df = new SimpleDateFormat("yyyy-mm-ddHH:mm:ss.SSSzzz");
private int skipCount;
private int indexCount;

@Setup(Level.Iteration)
public void createIndexer() throws Exception {
random = new Random();
registry = new SimpleMeterRegistry();
tempDirectory =
Files.createDirectories(
Paths.get("jmh-output", String.valueOf(random.nextInt(Integer.MAX_VALUE))));
logStore =
LuceneIndexStoreImpl.makeLogStore(
tempDirectory.toFile(),
commitInterval,
refreshInterval,
true,
SchemaAwareLogDocumentBuilderImpl.FieldConflictPolicy.CONVERT_VALUE_AND_DUPLICATE_FIELD,
registry);

apiLogFile = System.getProperty("jmh.api.log.file", "api_logs.txt");
reader = Files.newBufferedReader(Path.of(apiLogFile));
skipCount = 0;
indexCount = 0;
}

@TearDown(Level.Iteration)
public void tearDown() throws IOException {
Directory directory = logStore.getIndexWriter().getDirectory();
String[] segmentFiles = directory.listAll();
long indexedBytes = 0;
for (String segmentFile : segmentFiles) {
indexedBytes += directory.fileLength(segmentFile);
}
if (indexCount != 0) {
// Displaying indexCount only makes sense in measureAPILogIndexingSlingshotMode
System.out.println(
"Indexed = "
+ indexCount
+ " Skipped = "
+ skipCount
+ " Index size = "
+ FileUtils.byteCountToDisplaySize(indexedBytes));
} else {
System.out.println(
"Skipped = "
+ skipCount
+ " Index size = "
+ FileUtils.byteCountToDisplaySize(indexedBytes));
}
logStore.close();
try (Stream<Path> walk = Files.walk(tempDirectory)) {
walk.sorted(Comparator.reverseOrder()).map(Path::toFile).forEach(File::delete);
}
registry.close();
if (reader != null) {
reader.close();
}
}

@Benchmark
public void measureAPILogIndexing() throws IOException {
String line = reader.readLine();
if (line != null) {
// Work that ideally shouldn't count towards benchmark performance result
ConsumerRecord<String, byte[]> kafkaRecord = makeConsumerRecord(line);
if (kafkaRecord == null) {
// makeConsumerRecord will print why we skipped
return;
}
// Mimic LogMessageWriterImpl#insertRecord kinda without the chunk rollover logic
try {
LogMessage localLogMessage =
LogMessageWriterImpl.apiLogTransformer.toLogMessage(kafkaRecord).get(0);
logStore.addMessage(localLogMessage);
} catch (Exception e) {
System.out.println("skipping - cannot transform " + e);
skipCount++;
}
} else {
System.out.println("resetting - reach EOF");
reader = Files.newBufferedReader(Path.of(apiLogFile));
}
}

@Benchmark
public void measureAPILogIndexingSlingshotMode() throws IOException {
String line;
do {
line = reader.readLine();
if (line != null) {
// Work that ideally shouldn't count towards benchmark performance result
ConsumerRecord<String, byte[]> kafkaRecord = makeConsumerRecord(line);
if (kafkaRecord == null) {
// makeConsumerRecord will print why we skipped
continue;
}
// Mimic LogMessageWriterImpl#insertRecord kinda without the chunk rollover logic
try {
LogMessage localLogMessage =
LogMessageWriterImpl.apiLogTransformer.toLogMessage(kafkaRecord).get(0);
logStore.addMessage(localLogMessage);
indexCount++;
} catch (Exception e) {
System.out.println("skipping - cannot transform " + e);
}
}
} while (line != null);
}

public ConsumerRecord<String, byte[]> makeConsumerRecord(String line) {
try {
// get start of messageBody
int messageDivision = line.indexOf("{");

// Everything will there is metadata
String[] splitLine = line.substring(0, messageDivision - 1).split("\\s+");
String ts = splitLine[0] + splitLine[1] + splitLine[2] + splitLine[3];
long timestamp = df.parse(ts).toInstant().toEpochMilli();

String message = line.substring(messageDivision);
Murron.MurronMessage testMurronMsg =
Murron.MurronMessage.newBuilder()
.setMessage(ByteString.copyFrom((message).getBytes(StandardCharsets.UTF_8)))
.setType(splitLine[5])
.setHost(splitLine[4])
.setTimestamp(timestamp)
.build();
return new ConsumerRecord<>(
"testTopic",
1,
10,
0L,
TimestampType.CREATE_TIME,
0L,
0,
0,
"testKey",
testMurronMsg.toByteString().toByteArray());
} catch (Exception e) {
System.out.println("skipping - cannot parse input" + e);
skipCount++;
return null;
}
}
}
// package com.slack.kaldb;
//
// import com.google.protobuf.ByteString;
// import com.slack.kaldb.logstore.LogMessage;
// import com.slack.kaldb.logstore.LuceneIndexStoreImpl;
// import com.slack.kaldb.logstore.schema.SchemaAwareLogDocumentBuilderImpl;
// import com.slack.kaldb.writer.LogMessageWriterImpl;
// import com.slack.service.murron.Murron;
// import io.micrometer.core.instrument.MeterRegistry;
// import io.micrometer.core.instrument.simple.SimpleMeterRegistry;
// import java.io.BufferedReader;
// import java.io.File;
// import java.io.IOException;
// import java.nio.charset.StandardCharsets;
// import java.nio.file.Files;
// import java.nio.file.Path;
// import java.nio.file.Paths;
// import java.text.SimpleDateFormat;
// import java.time.Duration;
// import java.util.Comparator;
// import java.util.Random;
// import java.util.stream.Stream;
// import org.apache.commons.io.FileUtils;
// import org.apache.kafka.clients.consumer.ConsumerRecord;
// import org.apache.kafka.common.record.TimestampType;
// import org.apache.lucene.store.Directory;
// import org.openjdk.jmh.annotations.*;
//
// @State(Scope.Thread)
// public class IndexAPILog {
//
// private Random random;
// private final Duration commitInterval = Duration.ofSeconds(5 * 60);
// private final Duration refreshInterval = Duration.ofSeconds(5 * 60);
//
// private Path tempDirectory;
// private MeterRegistry registry;
// LuceneIndexStoreImpl logStore;
//
// private String apiLogFile;
// private BufferedReader reader;
// private static SimpleDateFormat df = new SimpleDateFormat("yyyy-mm-ddHH:mm:ss.SSSzzz");
// private int skipCount;
// private int indexCount;
//
// @Setup(Level.Iteration)
// public void createIndexer() throws Exception {
// random = new Random();
// registry = new SimpleMeterRegistry();
// tempDirectory =
// Files.createDirectories(
// Paths.get("jmh-output", String.valueOf(random.nextInt(Integer.MAX_VALUE))));
// logStore =
// LuceneIndexStoreImpl.makeLogStore(
// tempDirectory.toFile(),
// commitInterval,
// refreshInterval,
// true,
//
// SchemaAwareLogDocumentBuilderImpl.FieldConflictPolicy.CONVERT_VALUE_AND_DUPLICATE_FIELD,
// registry);
//
// apiLogFile = System.getProperty("jmh.api.log.file", "api_logs.txt");
// reader = Files.newBufferedReader(Path.of(apiLogFile));
// skipCount = 0;
// indexCount = 0;
// }
//
// @TearDown(Level.Iteration)
// public void tearDown() throws IOException {
// Directory directory = logStore.getIndexWriter().getDirectory();
// String[] segmentFiles = directory.listAll();
// long indexedBytes = 0;
// for (String segmentFile : segmentFiles) {
// indexedBytes += directory.fileLength(segmentFile);
// }
// if (indexCount != 0) {
// // Displaying indexCount only makes sense in measureAPILogIndexingSlingshotMode
// System.out.println(
// "Indexed = "
// + indexCount
// + " Skipped = "
// + skipCount
// + " Index size = "
// + FileUtils.byteCountToDisplaySize(indexedBytes));
// } else {
// System.out.println(
// "Skipped = "
// + skipCount
// + " Index size = "
// + FileUtils.byteCountToDisplaySize(indexedBytes));
// }
// logStore.close();
// try (Stream<Path> walk = Files.walk(tempDirectory)) {
// walk.sorted(Comparator.reverseOrder()).map(Path::toFile).forEach(File::delete);
// }
// registry.close();
// if (reader != null) {
// reader.close();
// }
// }
//
// @Benchmark
// public void measureAPILogIndexing() throws IOException {
// String line = reader.readLine();
// if (line != null) {
// // Work that ideally shouldn't count towards benchmark performance result
// ConsumerRecord<String, byte[]> kafkaRecord = makeConsumerRecord(line);
// if (kafkaRecord == null) {
// // makeConsumerRecord will print why we skipped
// return;
// }
// // Mimic LogMessageWriterImpl#insertRecord kinda without the chunk rollover logic
// try {
// LogMessage localLogMessage =
// LogMessageWriterImpl.apiLogTransformer.toLogMessage(kafkaRecord).get(0);
// logStore.addMessage(localLogMessage);
// } catch (Exception e) {
// System.out.println("skipping - cannot transform " + e);
// skipCount++;
// }
// } else {
// System.out.println("resetting - reach EOF");
// reader = Files.newBufferedReader(Path.of(apiLogFile));
// }
// }
//
// @Benchmark
// public void measureAPILogIndexingSlingshotMode() throws IOException {
// String line;
// do {
// line = reader.readLine();
// if (line != null) {
// // Work that ideally shouldn't count towards benchmark performance result
// ConsumerRecord<String, byte[]> kafkaRecord = makeConsumerRecord(line);
// if (kafkaRecord == null) {
// // makeConsumerRecord will print why we skipped
// continue;
// }
// // Mimic LogMessageWriterImpl#insertRecord kinda without the chunk rollover logic
// try {
// LogMessage localLogMessage =
// LogMessageWriterImpl.apiLogTransformer.toLogMessage(kafkaRecord).get(0);
// logStore.addMessage(localLogMessage);
// indexCount++;
// } catch (Exception e) {
// System.out.println("skipping - cannot transform " + e);
// }
// }
// } while (line != null);
// }
//
// public ConsumerRecord<String, byte[]> makeConsumerRecord(String line) {
// try {
// // get start of messageBody
// int messageDivision = line.indexOf("{");
//
// // Everything will there is metadata
// String[] splitLine = line.substring(0, messageDivision - 1).split("\\s+");
// String ts = splitLine[0] + splitLine[1] + splitLine[2] + splitLine[3];
// long timestamp = df.parse(ts).toInstant().toEpochMilli();
//
// String message = line.substring(messageDivision);
// Murron.MurronMessage testMurronMsg =
// Murron.MurronMessage.newBuilder()
// .setMessage(ByteString.copyFrom((message).getBytes(StandardCharsets.UTF_8)))
// .setType(splitLine[5])
// .setHost(splitLine[4])
// .setTimestamp(timestamp)
// .build();
// return new ConsumerRecord<>(
// "testTopic",
// 1,
// 10,
// 0L,
// TimestampType.CREATE_TIME,
// 0L,
// 0,
// 0,
// "testKey",
// testMurronMsg.toByteString().toByteArray());
// } catch (Exception e) {
// System.out.println("skipping - cannot parse input" + e);
// skipCount++;
// return null;
// }
// }
// }
Loading

0 comments on commit 8332866

Please sign in to comment.