Hello,
When we restart a node in ES 5.6, active shard percent is about 96%
We have noticed that spark jobs stop indexing in cluster as long as the value of active shard percent does not exceed 98%.
have you heard of this kind of behavior ?
Thanks for reply