Zero downtime

Our previous meetup was all about mappings and settings. We talked about achieving search and aggregation goals by using custom analyzers, multi fields and dynamic templates. The presentation is available at https://gist.github.com/abronner/dc983d3f2bfce183718f.

As long as the software is not released, there is no problem to modify the index and restart the cluster when needed. Once released, the system and the data should be available for all the users all the time. What do you do when you release a new version that modifies your index? When do you have to re-index your data? How do you deal with new incoming data during re-indexing? And how do you upgrade your entire cluster to a new version of elasticsearch?

The main presentation will talk about solving these problems with zero downtime of elasticsearch in production. There is an open slot for an additional presentation (please contact me), otherwise we will continue with an open discussion.