I've been testing ES v1.7.1 before upgrading all our clusters, and noticed a pretty serious bug when using it as tribe.
We are using tribe nodes to aggregate many downstream clusters. Over 90 clusters, 14K indices. Using v1.4.5 tribe node, I can see all downstream nodes and indices. Tested upgrading to 1.7.1 and it see only around 30% of downstream nodes, which also mean many indices are missing.
This is a major bug that is keeping us from upgrading. I've filed a bug with Elastic, just wanted to know if anyone else had similar experience.