We configured replication factor 2. With replication its coming around 300 tasks. Every hour 300 tasks are running. Problem is whenever middle manager is down, the ingestion is not smooth. Ideally ingestion should be smoother , druid should pickup the replicated node. But it is taking more than 15 mins to restore the ingestion. What could be the reason?
I just added tranquility parameter with the replication factor value. Other than that i didn’t change anything
Also another question:
Did all the replicated nodes data also written in hadoop? I am getting access violation error with replication enabled inside hadoop. This error ususally comes when same file is being accessd for write by two different hadoop clients