After some work producing the Debezium Cassandra Connector .jar file with dependencies, I can confirm that the connector works without any issues after considering the following:
1) log4j.properties file required:
2) Java command used:
3) I can confirm using GossipinPropertyFileSnitch cassandra cluster is posible (with a simple workaround)
4) CDC and CommitLog cassandra.yaml parameters:
Note: I am using a cassandra 3 node cluster using GossipingPropertyFileSnitch, ubuntu 16.04, Java 8, Cassandra 3.11 and Kafka 2.12-2.60
After some work producing the Debezium Cassandra Connector .jar file with dependencies, I can confirm that the connector works without any issues after considering the following:
1) log4j.properties file required:
the log4j.properties file I am using is the one that I found within the KAFKA tar (2.12-2.6.0)
2) Java command used:
in my case running it from my created folder $ /opt/debezium/plugin
java -cp "./*:./" io.debezium.connector.cassandra.CassandraConnectorTask config.properties -Dlog4j.configuration=file:///opt/debezium/plugin/log4j.properties
Keep in a directory together the debezium cassandra .jar, debezium cassandra .properties file (in my case named config.properties) and the log4j.properties file (found in kafka tar distribution). Also here I am keeping a slighly modified cassandra.yaml (in order to make it work with underlying cassandra cluster using the GossipingPropertyFileSnitch.
3) I can confirm using GossipinPropertyFileSnitch cassandra cluster is not posible if the debezium cassandra connector is not tricked using a modified and separated cassandra.yaml file. In my case what I did was to copy the cassandra.yaml from my cluster deployment using GossipinPropertyFileSnitch and modify the snitch parameter to "SimpleSnitch" and passing this file in the debezium cassandra .properties file using (cassandra.config=/path/to/modified/yaml). In my case I am keeping this modified cassandra.yaml file also in the same directory (/opt/debezium/plugin).
4) CDC and CommitLog cassandra.yaml parameters:
Follow cdc parameters as per documentation and if you want the connector to find the table(s) for snapshot, lower the following to see the kafka topic polulated. commitlog_segment=1 and commitlog_total_space_in_mb=2 will help to make work the connector with a cassandra with low traffic (edit it according to your needs).
Bug
Blocker
