I submitted https://github.com/onyx-platform/onyx-template/pull/77 with the java invocation required to get the template running, but I'm still a little confused on one point.
If my :in
for the job is a clojure.async channel, how do I go actually submitting data to this clojure.async channel?
The test (https://github.com/onyx-platform/onyx-template/blob/0.14.x/src/leiningen/new/onyx_app/test/onyx_app/jobs/basic_test.clj) uses (>!! in segment)
, but I don't have an nREPL into the peergroup in order to try & access this channel
Core async is more for unit/integration testing
Probably want to use kafka for a real job
Ahh. Yeah, my hunch is that the job isn't inter-actable and never completes (though the (exit 0 "Job Completed")
) in core.clj
led me to think otherwise
I did get a real Zookeeper instance running this morning -- already have the compiled Kafka. Despite the age, I've gotten great instruction out of https://dataissexy.wordpress.com/2016/07/31/using-onyx-template-to-craft-a-kafka-streaming-application-part-1-clojure-onyx-kafka-streaming-data/ so far
I'll continue onto Part 2 & go ahead with getting Kafka setup. This seems more sane than having the peer group initialize an HTTP server & pipe data into the async channel, which is a lot of throw-away work
onyx-kafka plugin has changed significantly -- the kafka-opts
for the task from the 2016 tutorial needed a lot of tweaking.
There was one key that's supposed to be allowed -- :kafka/commit-interval
-- and it's not marked as deprecated at https://github.com/onyx-platform/onyx-kafka/blob/b092ad0e96bd52bb38e9b447834d78c072a0fb47/src/onyx/kafka/information_model.cljc
However, keeping it in the task options was failing the Schema validation check every time.
I'm guessing this key was intended to be deprecated? If nobody in channel knows within a day or so, I'll file an issue.
created https://github.com/onyx-platform/onyx-kafka/issues/62