Hi
When your indexer goes down in cluster then CM try to spread out all buckets with your current indexers. This means that every indexers will divide those buckets which failed indexers have. In normal situation you shouldn't change SF&RF on this situation. Instead of you should fix that failed indexers and bring it back to cluster. After that cluster will rebalance buckets with all available indexers.
Basically indexer clusters works as it should.
But for that reason you should have enough spare space on those peers to manage situation when one (or even more, based on your RF&SF) can be down for some time. Other option is that your peers frozen some buckets to have enough space for normal operations. Of course this needs that you are using volumes with suitable max limit.
r. Ismo
