Verspätung is a small utility which aims to help identify delay of Kafka consumers.

Verspätung monitors the topics and their latest offsets by talking to Kafka, it will also keep track of how far along consumers are by monitoring the offsets that they have committed to Zookeeper. Using both of these pieces of information, Verspätung computs the delta for each of the consumer groups and reports it to statsd.

Using

% java -jar verspaetung-*-all.jar --help
usage: verspaetung
 -d,--delay <DELAY>          Seconds to delay between reporting metrics to
                             the metrics receiver (defaults: 5s)
 -H,--statsd-host <STATSD>   Hostname for a statsd instance (defaults to
                             localhost)
 -n,--dry-run                Disable reporting to a statsd host
 -p,--statsd-port <PORT>     Port for the statsd instance (defaults to
                             8125)
    --prefix <PREFIX>        Prefix all metrics with PREFIX before they're
                             reported (e.g. PREFIX.verspaetung.mytopic)
 -s,--storm                  Watch Storm KafkaSpout offsets (under
                             /kafka_spout)
 -x,--exclude <EXCLUDES>     Regular expression for consumer groups to
                             exclude from reporting (can be declared
                             multiple times)
 -z,--zookeeper <HOSTS>      Comma separated list of Zookeeper hosts (e.g.
                             localhost:2181)

Running Verspätung is rather easy, by default the daemon will monitor the standard Kafka high-level consumer offset path of /consumers and start reporting deltas automatically.