We are using Apache Druid version 0.19.0 in cluster mode.
There are 14 nodes in this cluster,
2 Master nodes (Coordinator + Overload process)
2 Query nodes (Broker + Router process)
10 Data nodes (Historical + Middle manager process)
I have gone through the details given in following link -
In current system there are multiple data source ingesting real time data through Kafka stream.
I have downloaded the 0.20.1 version and copied all the files from
/apache-druid-0.19.0/conf/druid/cluster/ to /apache-druid-0.20.1/conf/druid/cluster/
Also I have linked the /apache-druid-0.20.1**/var** to current var location.
I wanted to know how can we upgrade to new version without any downtime or data loss. What steps should I need to take ?
- Should I need to stop all the 14 servers processes first and then start with new configs?
- Should I need to stop 1 out of 10 data server and then start 1 data server with new config and repeat the same process for all 10 data server?
- Since same server is used for Historical and Middle manger then how can we start only historical then overload then middle manage as per given link?
Hope you understand my confusion. Can you please suggest the approach.
Thanks & Regards