Kafka JDBC source connector time stamp mode failin

问题:

I tried to set up a database with two tables in sqlite. Once of my table is having a timestamp column . I am trying to implement timestamp mode to capture incremental changes in the DB. Kafka connect is failing with the below error:

 ERROR Failed to get current time from DB using Sqlite and query 'SELECT 
CURRENT_TIMESTAMP' 
(io.confluent.connect.jdbc.dialect.SqliteDatabaseDialect:471)
java.sql.SQLException: Error parsing time stamp

Caused by: java.text.ParseException: Unparseable date: "2019-02-05 02:05:29" 
does not match (\p{Nd}++)\Q-\E(\p{Nd}++)\Q-\E(\p{Nd}++)\Q 
\E(\p{Nd}++)\Q:\E(\p{Nd}++)\Q:\E(\p{Nd}++)\Q.\E(\p{Nd}++)

Many thanks for the help

Config:

name=test-query-sqlite-jdbc-autoincrement 
connector.class=io.confluent.connect.jdbc.JdbcSourceConnector 
tasks.max=1 
connection.url=jdbc:sqlite:employee.db 
query=SELECT users.id, users.name, transactions.timestamp, transactions.payment_type FROM users JOIN transactions ON (users.id = transactions.user_id) 
mode=timestamp 
timestamp.column.name=timestamp 
topic.prefix=test-joined

DDL:

CREATE TABLE transactions(id integer primary key not null,
                          payment_type text not null,
                          timestamp DATETIME DEFAULT(STRFTIME('%Y-%m-%d %H:%M:%f', 'NOW')),
                          user_id int not null, 
                          constraint fk foreign key(user_id) references users(id)
); 

CREATE TABLE users (id integer primary key not null,name text not null);

回答1:

The kafka connect jdbc connector easily detects the changes in the timestamp, if the values of the 'timestamp' column are in the format of the 'UNIX timestamp'.

sqlite> CREATE TABLE transact(timestamp TIMESTAMP DEFAULT (STRFTIME('%s', 'now')) not null,
   ...> id integer primary key not null,
   ...> payment_type text not null);
sqlite>

The values can be inserted as:

sqlite> INSERT INTO transact(timestamp,payment_type,id) VALUES (STRFTIME('%s', 'now'),'cash',1);

The timestamp related changes are then detected by the kafka jdbc source connector and the same can be consumed as follows:

kafka-console-consumer  --bootstrap-server localhost:9092 --topic jdbc-transact --from-beginning
{"timestamp":1562321516,"id":2,"payment_type":"card"}
{"timestamp":1562321790,"id":1,"payment_type":"online"}

回答2:

I've reproduced this, and it is already logged as an issue for the JDBC Source connector. You can monitor it here: https://github.com/confluentinc/kafka-connect-jdbc/issues/219