[jira] [Created] (LOG4J2-2678) Add LogEvent timestamp to ProducerRecord in KafkaAppender

["Federico D'Ambrosio (Jira)" <ji...@apache.org>]

Federico D'Ambrosio created LOG4J2-2678:
-------------------------------------------

             Summary: Add LogEvent timestamp to ProducerRecord in KafkaAppender
                 Key: LOG4J2-2678
                 URL: https://issues.apache.org/jira/browse/LOG4J2-2678
             Project: Log4j 2
          Issue Type: New Feature
          Components: Appenders
            Reporter: Federico D'Ambrosio


Just like I anticipated in the Mailing List, I'm creating this issue to add the possibility to send the LogEvent time as a timestamp for the record when using the log4j KafkaAppender.

As far as I've seen in the source code, the message is sent by KafkaManager:


{code:java}
private void tryAppend(final LogEvent event) throws ExecutionException, InterruptedException, TimeoutException {
    final Layout<? extends Serializable> layout = getLayout();
    byte[] data;
    if (layout instanceof SerializedLayout) {
       final byte[] header = layout.getHeader();
       final byte[] body = layout.toByteArray(event);
       data = new byte[header.length + body.length];
       System.arraycopy(header, 0, data, 0, header.length);
       System.arraycopy(body, 0, data, header.length, body.length);
    } else {
     data = layout.toByteArray(event);
   }
  manager.send(data); //manager.send(data, event.getTimeMillis())
}
{code}


with manager.send() implemented this way, with highlighted the creation of the ProducerRecord:

{code:java}
public void send(final byte[] msg) throws ExecutionException, InterruptedException, TimeoutException {
    if (producer != null) {
        byte[] newKey = null;

        if(key != null && key.contains("${")) {
             newKey = getLoggerContext().getConfiguration().getStrSubstitutor().replace(key).getBytes(StandardCharsets.UTF_8);
        } else if (key != null) {
             newKey = key.getBytes(StandardCharsets.UTF_8);
        }

    final ProducerRecord<byte[], byte[]> newRecord = new ProducerRecord<>(topic, newKey, msg);
    if (syncSend) {
        final Future<RecordMetadata> response = producer.send(newRecord);
        response.get(timeoutMillis, TimeUnit.MILLISECONDS);
    } else {
        producer.send(newRecord, new Callback() {
             @Override
             public void onCompletion(final RecordMetadata metadata, final Exception e) {
                 if (e != null) {
                     LOGGER.error("Unable to write to Kafka in appender [" + getName() + "]", e);
             }
         }});
   }
}
}
{code}


Now, ProducerRecord has the additional parameters, in particular, I'm looking at: [this constructor|https://kafka.apache.org/10/javadoc/org/apache/kafka/clients/producer/ProducerRecord.html#ProducerRecord-java.lang.String-java.lang.Integer-java.lang.Long-K-V- ]


{code:java}
public ProducerRecord(java.lang.String topic,
                      java.lang.Integer partition,
                      java.lang.Long timestamp,
                      K key,
                      V value)
{code}


which would allow us to set the timestamp as {{LogEvent#getTimeMillis()}}, but would force us to also input the partition where the record should be sent. Still, the logic behind the partitioning within the KafkaProducer is so that if {{partition}} is null, then the defined partitioner will be used (DefaultPartitioner or the one defined by the '{{partitioner.class}}' property), so, we could simply assign it as null.

In terms of interface, we could add a single flag in the KafkaAppender definition, something like:


{code:xml}
<Kafka name="kafka-appender" topic="topic" timestamp="true"> </Kafka>
{code}


If the 'timestamp' flag is false, then the record would be sent with the timestamp parameter of the ProducerRecord as null, leaving the behaviour as it is right now.




--
This message was sent by Atlassian Jira
(v8.3.2#803003)