/**
* This file is part of Graylog.
*
* Graylog is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* Graylog is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with Graylog. If not, see <http://www.gnu.org/licenses/>.
*/
package org.graylog2.outputs;
import com.codahale.metrics.Histogram;
import com.codahale.metrics.Meter;
import com.codahale.metrics.MetricRegistry;
import com.codahale.metrics.Timer;
import com.google.common.collect.Lists;
import com.google.common.collect.Maps;
import com.google.inject.assistedinject.Assisted;
import com.google.inject.assistedinject.AssistedInject;
import org.graylog2.indexer.IndexSet;
import org.graylog2.indexer.cluster.Cluster;
import org.graylog2.indexer.messages.Messages;
import org.graylog2.plugin.Message;
import org.graylog2.plugin.configuration.Configuration;
import org.graylog2.plugin.streams.Stream;
import org.graylog2.shared.journal.Journal;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import javax.inject.Inject;
import java.util.List;
import java.util.Map;
import java.util.concurrent.TimeoutException;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.atomic.AtomicLong;
import static com.codahale.metrics.MetricRegistry.name;
import static java.util.concurrent.TimeUnit.NANOSECONDS;
// Singleton class
public class BlockingBatchedESOutput extends ElasticSearchOutput {
private static final Logger log = LoggerFactory.getLogger(BlockingBatchedESOutput.class);
private final Cluster cluster;
private final int maxBufferSize;
private final Timer processTime;
private final Histogram batchSize;
private final Meter bufferFlushes;
private final Meter bufferFlushesRequested;
private volatile List<Map.Entry<IndexSet, Message>> buffer;
private static final AtomicInteger activeFlushThreads = new AtomicInteger(0);
private final AtomicLong lastFlushTime = new AtomicLong();
private final int outputFlushInterval;
@AssistedInject
public BlockingBatchedESOutput(MetricRegistry metricRegistry,
Messages messages,
Cluster cluster,
org.graylog2.Configuration serverConfiguration,
Journal journal,
@Assisted Stream stream,
@Assisted Configuration configuration) {
this(metricRegistry, messages, cluster, serverConfiguration, journal);
}
@Inject
public BlockingBatchedESOutput(MetricRegistry metricRegistry,
Messages messages,
Cluster cluster,
org.graylog2.Configuration serverConfiguration,
Journal journal) {
super(metricRegistry, messages, journal);
this.cluster = cluster;
this.maxBufferSize = serverConfiguration.getOutputBatchSize();
outputFlushInterval = serverConfiguration.getOutputFlushInterval();
this.processTime = metricRegistry.timer(name(this.getClass(), "processTime"));
this.batchSize = metricRegistry.histogram(name(this.getClass(), "batchSize"));
this.bufferFlushes = metricRegistry.meter(name(this.getClass(), "bufferFlushes"));
this.bufferFlushesRequested = metricRegistry.meter(name(this.getClass(), "bufferFlushesRequested"));
buffer = Lists.newArrayListWithCapacity(maxBufferSize);
}
@Override
public void write(Message message) throws Exception {
for (IndexSet indexSet : message.getIndexSets()) {
writeMessageEntry(Maps.immutableEntry(indexSet, message));
}
}
public void writeMessageEntry(Map.Entry<IndexSet, Message> entry) throws Exception {
List<Map.Entry<IndexSet, Message>> flushBatch = null;
synchronized (this) {
buffer.add(entry);
if (buffer.size() >= maxBufferSize) {
flushBatch = buffer;
buffer = Lists.newArrayListWithCapacity(maxBufferSize);
}
}
// if the current thread found it had to flush any messages, it does so but blocks.
// this ensures we don't flush more than 'processorCount' in parallel.
// TODO this will still be time limited by the OutputBufferProcessor and thus be called more often than it should
if (flushBatch != null) {
flush(flushBatch);
}
}
private void flush(List<Map.Entry<IndexSet, Message>> messages) {
if (!cluster.isConnected() || !cluster.isDeflectorHealthy()) {
try {
cluster.waitForConnectedAndDeflectorHealthy();
} catch (TimeoutException | InterruptedException e) {
log.warn("Error while waiting for healthy Elasticsearch cluster. Not flushing.", e);
return;
}
}
// never try to flush an empty buffer
if (messages.size() == 0) {
return;
}
log.debug("Starting flushing {} messages, flush threads active {}",
messages.size(),
activeFlushThreads.incrementAndGet());
try (Timer.Context ignored = processTime.time()) {
lastFlushTime.set(System.nanoTime());
writeMessageEntries(messages);
batchSize.update(messages.size());
bufferFlushes.mark();
} catch (Exception e) {
log.error("Unable to flush message buffer", e);
}
activeFlushThreads.decrementAndGet();
log.debug("Flushing {} messages completed", messages.size());
}
public void forceFlushIfTimedout() {
if (!cluster.isConnected() || !cluster.isDeflectorHealthy()) {
// do not actually try to flush, because that will block until the cluster comes back.
// simply check and return.
log.debug("Cluster unavailable, but not blocking for periodic flush attempt. This will try again.");
return;
}
// if we shouldn't flush at all based on the last flush time, no need to synchronize on this.
if (lastFlushTime.get() != 0 &&
outputFlushInterval > NANOSECONDS.toSeconds(System.nanoTime() - lastFlushTime.get())) {
return;
}
// flip buffer quickly and initiate flush
final List<Map.Entry<IndexSet, Message>> flushBatch;
synchronized (this) {
flushBatch = buffer;
buffer = Lists.newArrayListWithCapacity(maxBufferSize);
}
if (flushBatch != null) {
bufferFlushesRequested.mark();
flush(flushBatch);
}
}
public interface Factory extends ElasticSearchOutput.Factory {
}
public static class Config extends ElasticSearchOutput.Config {
}
public static class Descriptor extends ElasticSearchOutput.Descriptor {
public Descriptor() {
super("Blocking Batched Elasticsearch Output", false, "", "Elasticsearch Output with Batching (blocking)");
}
}
}