NullPointerException when connecting Confluent Kafka and InfluxDB - apache-nifi

Question

I'm trying to use the Confluent InfluxDB Sink Connector to get data from a kafka topic into my InfluxDB.
Firstly, I transmit data to kafka topic from a log file by using nifi, and it works well. The kafka topic get the data, like below:
{
"topic": "testDB5",
"key": null,
"value": {
"timestamp": "2019-03-20 01:24:29,461",
"measurement": "INFO",
"thread": "NiFi Web Server-795",
"class": "org.apache.nifi.web.filter.RequestLogger",
"message": "Attempting request for (anonymous)
},
"partition": 0,
"offset": 0
}
Then, I create InfluxDB sink connector through the Kafka Connect UI , and I get the following exception:
org.apache.kafka.connect.errors.ConnectException: Exiting WorkerSinkTask due to unrecoverable exception.
at org.apache.kafka.connect.runtime.WorkerSinkTask.deliverMessages(WorkerSinkTask.java:587)
at org.apache.kafka.connect.runtime.WorkerSinkTask.poll(WorkerSinkTask.java:323)
at org.apache.kafka.connect.runtime.WorkerSinkTask.iteration(WorkerSinkTask.java:226)
at org.apache.kafka.connect.runtime.WorkerSinkTask.execute(WorkerSinkTask.java:194)
at org.apache.kafka.connect.runtime.WorkerTask.doRun(WorkerTask.java:175)
at org.apache.kafka.connect.runtime.WorkerTask.run(WorkerTask.java:219)
at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
at java.util.concurrent.FutureTask.run(FutureTask.java:266)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
Caused by: java.lang.NullPointerException
at io.confluent.influxdb.InfluxDBSinkTask.put(InfluxDBSinkTask.java:140)
at org.apache.kafka.connect.runtime.WorkerSinkTask.deliverMessages(WorkerSinkTask.java:565)
... 10 more
But if I manually input data to another topic testDB1 by using
./bin/kafka-avro-console-producer --broker-list localhost:9092 --topic testDB1 --property value.schema='{"type":"record","name":"myrecord","fields":[{"name":"measurement","type":"string"},{"name":"timestamp","type":"string"}]}'
It works, my influxDB can get the data.
Here is the connect configuration:
connector.class=io.confluent.influxdb.InfluxDBSinkConnector
influxdb.url=http://myurl
tasks.max=1
topics=testDB5
the configuration of connecting topic testDB1 is the same except topics name.
Is there any problems in nifi ? But it can transmit data to topic well.

Answer 1

When you use Avro with Kafka Connect, the Avro deserialiser expects the data to have been serialised using the Avro serialiser. This is what the kafak-avro-console-producer uses, which is why your pipeline works when you use that.
This article gives a good background to Avro and the Schema Registry. See also Kafka Connect Deep Dive – Converters and Serialization Explained.
I'm not familiar with Nifi, but looking at the documentation it seems that AvroRecordSetWriter has the option to use Confluent Schema Registry. At a guess you'll also want to set Schema Write Strategy to Confluent Schema Registry Reference.
Once you can consume data from your topic with kafka-avro-console-consumer then you know that it is correctly serialised and will work with your Kafka Connect sink.

Answer 2

I found the reason. It's because in Nifi, I used PublishKafka_0_10 to publish the data to Kafka topic, but its version is to low!
When I make a query in ksql, it says that
Input record ConsumerRecord(..data..) has invalid (negative) timestamp.
Possibly because a pre-0.10 producer client was used to write this record to Kafka without embedding a timestamp,
or because the input topic was created before upgrading the Kafka cluster to 0.10+. Use a different TimestampExtractor to process this data.
So, I change it to PublishKafka_1_0 , and start again, and it works! My influxDB can get the data. I'm speechless.
And thanks Robin Moffatt for the reply, its very helpful to me.