I was in the middle of massive delete of the documents (routing was always to the same shard) when I started to get this exception on the bulk operation :
RemoteTransportException[[Jackdaw][inet[/10.0.XX.YY:9300]][indices:data/write/bulk[s]]]; nested: EngineClosedException[[myindex] CurrentState[CLOSED] ];
Where cluster was green, replica shard was started and running, and prime shared was marked as started but without any information on what is going on it.
Search on documents placed on that shard was working fine.
What could be an issue here and how can I start this prime shard again.