The twitter connector pom builds a fat jar with all dependencies. You need to add this to the classpath before you start Connect. This is what the Confluent scripts are doing.
Regards Andrew ________________________________ From: Ewen Cheslack-Postava<mailto:[email protected]> Sent: 14/06/2016 07:35 To: [email protected]<mailto:[email protected]> Subject: Re: Running kafka connector application Kanagha, I'm not sure about that particular connector, but normally the build script would provide support for collecting the necessary dependencies. Then all you need to do is add something like /path/to/connector-and-deps/* to your classpath and it shouldn't be affected by versions in the pom.xml, you'd just rebuild to pull in the new dependencies. For example, Confluent's connectors are all setup so if you build with mvn package, it'll include directories with all the dependency jars included. -Ewen On Sun, Jun 12, 2016 at 2:36 PM, Kanagha <[email protected]> wrote: > Hi, > > > I'm running the TwitterProducer task as per > https://github.com/Eneco/kafka-connect-twitter > > connect-standalone <path>/connect-source-standalone.properties > <path>/twitter-source.properties > > I see that I have to set the CLASSPATH to include all the dependent jars > that the target connector jar is dependent on. This method wouldn't be > robust if the version changes in pom.xml > > Is there an easier approach to run the standlone connect application? > > > Thanks > Kanagha > > > > > > Kanagha > -- Thanks, Ewen
