kafka connect hdfs flatten array records - apache-kafka

We are experimenting with Kafka Connect HDFS and have source data which contains an array of avro records (referred to as container record below) which are pushed to topics.
The plan was to use kafka-connect hdfs to flatten the array in the container record into individual records and write to hdfs as avro and then commit the actual container record.
However it seems that kafka-connect-hdfs does not support this out of the box.
In attempt to navigate the source we found that there is a tight mapping of the topic names and offset management and attempting to transform 1 sinkRecod to multiple sinkRecords deviates from expected behavior.
so wanted to verify if 1 to n transformation possible ? if so how ?

Kafka Connect supports Single Message Transforms (SMT) which is the area of functionality, if any, in Kafka Connect that would do what you want. However IIRC SMT are only for 1:1 input/output record processing, not 1:n - that is, I don't think (but am willing to be corrected) that you can generate additional output records from a single input record.
You probably want to look at Kafka Streams which would give you the full capabilities of doing this. You'd write a streams app which would subscribe to the source topic, do the necessary array processing, and write it back to a new Kafka topic. The new Kafka topic would then be the source topic for your Kafka Connect HDFS sink.

Related

How to monitor 'bad' messages written to kafka topic with no schema

I use Kafka Connect to take data from RabbitMQ into kafka topic. The data comes without schema so in order to associate schema I use ksql stream. On top of the stream I create a new topic that now has a defined schema. At the end I take the data to BQ database. My question is how do I monitor messages that have not passed the stream stage? in this way, do i support schema evolution? and if not, how can use the schema registry functionality?
Thanks
use Kafka Connect to take data ... data comes without schema
I'm not familiar specifically with Rabbitmq connector, but if you use the Confluent converter classes that do use schemas, then it would have one, although maybe only a string or bytes schema
If ksql is consuming the non-schema topic, then there's a consumer group associated with that process. You can monitor its lag to know how many messages have not yet been processed by ksql. If ksql is unable to parse a message because it's "bad", then I assume it's either skipped or the stream stops consuming completely; this is likely configurable
If you've set the output topic format to Avro, for example, then the schema will automatically be registered to the Registry. There will be no evolution until you modify the fields of the stream

Sink events of a same Kafka topic into multiple paths in GCS

I am using the Schema Registry with RecordNameStrategy naming policy so I have events with totally different avro schemas into the same Kafka topic.
I am doing that as I want to group logically related events that may have different data structures under a same topic to keep order for these data.
For instance:
user_created event and user_mail_confirmed event might have different schemas but it's important to keep them into a same topic partition to guarantee order for consumers.
I am trying to sink these data, coming from a single topic, into GCS in multiple paths (one path for each schema)
Does someone know if the Confluent Kafka connect GCS Sink connector (or any other connector) provide us with that feature please ?
I haven't used GCS connector, but I suppose that this is not possible with Confluent connectors in general.
You should probably copy your source topic with different data structures to a new set of topics, where data have common data structure. This is possible with ksqlDB (check an example) or Kafka Streams application. Then, you can create connectors for these topics.
Alternatively, you can use RegexRouter transformation with a set of predicates based on the message headers.

MQTT topics and kafka topics mapping

I have started to learn about MQTT as I have a use case in telematics in my current organisation. I would like to integrate MQTT broker ( mosquitto ) messages to my kafka.
Since every vehicle is sending the data in its own topic in MQTT broker within a single organisation, I would like to push all this data in kafka. Now I know it is not advisable to create so many topics in kafka ( more than a million ). Also I would like not like to save all the vehicles data in one kafka topic as I would like to later put all this data in S3, differentiated via vehicle id.
How can I achieve this without making so many topics in kafka. One way is the consumer of kafka segregate the events and put in s3 but I believe there will be a lot of small files in S3.
Generally, if you have the same logical entity you would use the same topic.
You can use the MQTT plugin for Kafka Connect to stream the data from MQTT into Kafka, and Kafka Connect's Single Message Transform RegexRouter to modify the topic name to which messages are written, and other SMT to modify the message key. That way you get all the messages in one topic, partitioned based on the vehicle id. That's probably the best way to store it.
From there, you can use the data however you want. When it comes to stream it to S3, you can use Kafka Connect S3 sink and as cricket_007 mentioned partition the data by time if it's just volume you're worried about. If you want to route the messages to different buckets or areas of the same bucket you could use a stream processing (e.g. Kafka Streams / ksqlDB) to pre-process the topic to populate others.
See here for an example of the MQTT connector.

Where to run the processing code in Kafka?

I am trying to setup a data pipeline using Kafka.
Data go in (with producers), get processed, enriched and cleaned and move out to different databases or storage (with consumers or Kafka connect).
But where do you run the actual pipeline processing code to enrich and clean the data? Should it be part of the producers or the consumers? I think I missed something.
In the use case of a data pipeline the Kafka clients could serve both as a consumer and producer.
For example, if you have raw data being streamed into ClientA where it is being cleaned before being passed to ClientB for enrichment then ClientA is serving as a consumer (listening to a topic for raw data) and a producer (publishing cleaned data to a topic).
Where you draw those boundaries is a separate question.
It can be part of either producer or consumer.
Or you could setup an environment dedicated to something like Kafka Streams processes or a KSQL cluster
It is possible either ways.Consider all possible options , choose an option which suits you best. Lets assume you have a source, raw data in csv or some DB(Oracle) and you want to do your ETL stuff and load it back to some different datastores
1) Use kafka connect to produce your data to kafka topics.
Have a consumer which would consume off of these topics(could Kstreams, Ksql or Akka, Spark).
Produce back to a kafka topic for further use or some datastore, any sink basically
This has the benefit of ingesting your data with little or no code using kafka connect as it is easy to set up kafka connect source producers.
2) Write custom producers, do your transformations in producers before
writing to kafka topic or directly to a sink unless you want to reuse this produced data
for some further processing.
Read from kafka topic and do some further processing and write it back to persistent store.
It all boils down to your design choice, the thoughput you need from the system, how complicated your data structure is.

Using kafka streams to create a table based on elasticsearch events

Is it possible to use Kafka streaming to create a pipeline that reads JSON from a Kafka topic and then do some logic with them and send the results to another Kafka topic or something else?
For example, I populate my topic using logs from elasticsearch. That is pretty easy using a simple logstash pipeline.
Once I have my logs in the kafka topic, I want to extract some pieces of information from the log and put them in a sort of "table" with N column(is Kafka capable of this?) and then put the table somewhere else (another topic or a db).
I didn't find any example that satisfies my criteria.
thanks
Yes, it's possible.
There is no concept of columns in kafka or kafka-streams. However, you typically just define a plain old java object of your choice, with the fields that your want (fields being the equivalent of columns in this case). You produce the output in that format to an output topic (using an appropriately chosen serializer). Finally, if you want to store the result in a relational database, you map the fields into columns, typically using a kafka connect jdbc sink:
http://docs.confluent.io/current/connect/connect-jdbc/docs/sink_connector.html