I've been trying to implement a 'reindex complete' notification system using Watcher APIs. Reindex is triggered with
wait_for_completion=true query parameter, so the reindex task is saved to
.tasks index, that is queried every minute searching for completed reindex tasks.
I've already implemented the most of the logic but I can't find example of the response in case of failed reindex. I have also tried simulating this case by turning off one of the nodes, but with no luck.
If anyone could post the example of a failed reindex task document, it would be much appriciated.