HTTP-based Elasticsearch destination

Add a native, HTTP based destination Elasticsearch, called elasticsearch-http(), as an alternative of the Java-based destination. This destination does not requires Java, and has smaller resource requirements. Eventually, we expect this destination to replace the Java implementation. For details, see "elasticsearch-http: Sending messages to Elasticsearch HTTP Bulk API" in the Administration Guide.

C-based Kafka destination

Add a native destination based on the librdkafka library, as an alternative of the Java-based destination. This destination does not requires Java, and has smaller resource requirements. Eventually, we expect this destination to replace the Java implementation. Note that this destination requires a very recent version of librdkafka. For details, see "kafka: Publishing messages to Apache Kafka (C implementation)" in the Administration Guide.

New parser

The CheckPoint LogExporter parser can parse CheckPoint log messages, and join the related multiline log messages into a single log message. For details, see Administration Guide.

New template functions

The $(implode) and $(explode) template functions allow you to split and join strings based on a simple separator. The exploded array is represented as a syslog-ng list that can be manipulated with the $(list-*) template functions. For details, see "Template functions of syslog-ng OSE" in the Administration Guide.

Enhancements

• The amqp() destination now supports the heartbeat and the external authentication mechanisms of AMQP. For details, see "amqp: Publishing messages using AMQP" in the Administration Guide.

• The graylog2() destination now supports the TLS and the UDP transport mechanisms. For details, see "Sending logs to Graylog" in the Administration Guide.

• The apache-accesslog-parser() now supports vhost:port as the first field of the log message in common and combined log formats.

• The grouping-by() parser now can order the messages of the context by using the sort-key() option.

• You can automatically determine the timezone of an incoming log message if the incoming stream is close to real time, and the timezone information is missing from the timestamp. You can enable this feature by using the flags(guess-timezone) option in sources and the date-parser().

• Until now, the syslog() source automatically closed the connection to the sender if it received a message that was longer than the value of the log-msg-size() option. From now on, if you set the trim-large-messages() option to yes, syslog-ng OSE only truncates the message to log-msg-size(), and sends a log message to the internal() source as notification about the truncation.

• You can now set the maximum size of spoofed datagrams in udp() destinations by using the spoof-source-max-msglen() option. Earlier, this size was hard-wired to 1024 bytes.

• The --omit-empty-values option for destinations and template functions using value-pairs allows you to skip empty value-pairs in the output.

• When parsing and classifying log messages with db-parser(), syslog-ng OSE so far used the value of the $PROGRAM field to determine which patterns to match to the incoming message. Now you can use program-template() option to customize the value which selects the patterns to use. For details, see "Using pattern databases" in the Administration Guide.

• The pdbtool merge tool can now sort the resulting database using the sort option. For details, see the pdbtool manual page.