about druid cluster questions

1:how to set realtime node HA?

2:how to set coordinator node HA?

3:how to set historical node HA?

4:how to set broker node HA?

5:about druid.discovery.curator.path=/data0/druid/zookeeper/prod/discovery how to make diffent node discovery other node???

i don’t use indexing node to cluster.

  1. Realtime nodes only have very limited HA ability. You need to be reading from Kafka without any Druid-side partitioning. By this I mean you have all of your Kafka partitions going into a single Druid partition. In this case, you can start up multiple realtime nodes with different consumer groups and things should generally work. But if you want to create multiple Druid partitions (perhaps because you have a large stream) then I would encourage you to look at tranquility + the indexing service (https://github.com/druid-io/tranquility). It can offer replicated realtime ingestion at any scale.

  2. Just start up multiple coordinators. One will be leader at any given time, and the others will be on standby. They handle their own leader election.

  3. Just start up multiple historical nodes, and make sure you set replication to 2 or higher.

  4. Just start up multiple brokers. You can put them all behind a load balancer.

  5. Sorry, I’m not sure I understand what you’re asking…