Monthly report for April 2018

Kibana

I spent some time working with Kibana and trying to understand some data inconsistency when using the Elasticsearch has_parent query. For a given query using it, it was obvious not every documents were returned. Using the same datsaset on a different Elasticsearch instance, I still had inconsistency though the results were different.

After properly reading the dozen pages of documentation about the join queries, I found something interesting in the definitive guide:

Elasticsearch maintains a map of which parents are associated with which children. It is thanks to this map that query-time joins are fast, but it does place a limitation on the parent-child relationship: the parent document and all of its children must live on the same shard.

This same shard limitation is what I was looking for: when I inserted the dataset, I had let Elasticsearch handle sharding. Solving my problem was rather simple: I created a new index and reindexed my data using the reindex API with a specific _routing field parameter. a inserted the dataset again using a custom routing field.

And inconsistency was gone.

Web landscape

I reviewed and tested a couple of handfuls of the mainstream tools in the JavaScript ecosystem. I was lagging a bit from the hype state of the art and took some time to catch up.

That reminds me about the post The Front-End Spectrum by Jeff Pelletier, and how much it remains relevant even after 3 years (I could bet the same article could have been written in 2011). Anyway, tools change, goal remains the same: deliver reliable and high quality software.