Kill a Realtime node during ingestion

I have multiple Realtime nodes consuming data from different Kafka partitions (using linear shardSpec). I understand that when a Realtime node is killed, other Realtime nodes re-distribute the load/partitions.

  1. What is the proper way of killing a Realtime node so that data which that node had consumed is persisted to deep storage before the node is actually killed?

  2. Can I just do "kill -9 "?

Hi Saksham,

I’m not sure, but i think that you can’t force the persist to deep storage. I think that if you kill the realtime node, it loses the data that it has in RAM. The realtime node writes data to tmp directory and it can recover data from this directory if it restart, but if you decide stop it forever you must wait the it do the deep storage when end the segment.

But I’am not sure … if there is any way to force the persist to deep storage.