GSoC: The second evaluation phase

It’s week 9 of Google Summer of Code and we’ve officially reached the second evaluation phase, which means we only have about four weeks left until our project comes to an end!

So far so good! The amount I’ve learnt so far and the amount I still have to learn is definitely exciting, and I’m very grateful to be able to work on this project. This last month has been heavily focused on research, a lot more so than I initally anticipated. We’ve also been attempting to get more involved in the community, especially since one of our mentors was away.

We faced a few blockers and experienced our fair share of ups and downs but we’ve still managed to make a fair bit of progress, with support from the community. All of the work so far has been leading up to the huge last month of work that awaits us.

• We created a new Kafka module in the UPS for Kafka related configs and tests (#870)

• We’ve updated the README with instructions for running the Kafka cluster on Openshift using templates by the enmasse project (#862)

• We’ve been playing around with Structure101 and SonarQube to analyse the UPS codebase. We’ve managed to get SonarQube running with JaCoCo code coverage reports and that’s all described a bit more in depth in my last blog post. (#865, #866)

• We got the CDI library installed and the first producer replaced. Getting the consumer injection going proved to be a slightly more difficult feat. We encountered multiple exceptions on redeployment of the UPS (see the thread on the mailing list, and these jiras: AGPUSH-2139, AGPUSH-2140, AGPUSH-2145). Polina got to work and spent loads of time working with the library to get this resolved, so we’re hoping to have this up and running in the very near future.

• We’ve been researching Kafka Security and the Kafka Streams API. I’ve worked on a few small Streams examples which I’ll hopefully be posting a blog about soon. As for security we started outlining the major points that will need to be worked on but we’ve decided to prioritize other tasks for the time being.

• We’ve made a few small Jira’s for minor UPS improvements which we’ve put under the gsoc_2017 and good_first_bug labels.

The biggest goal we’ve set for ourselves in our 4 week plan is to replace the entire JMS messaging system with Kafka producers and consumers, hopefully leveraging streams where possible. I’ve started work on this this week and so far, so good! Things seem to be going moderately well for the time being. I’ll hopefully be posting weekly updates from now on.