I want to set up a solr cloud using 3 different machines. All the examples i came across are advising me to download zookeeper as well.
However Solr comes with an inbuilt Zookeeper. Why cannot we use that?
Right now the embedded zookeeper is supposed to be mostly for testing/developing. Reasons being that it is less battle tested in production and that if Solr process goes down, zk goes down too, making the cluster less resilient than a separate process.
That being said, there are some voices saying that setting up a separate zk ensemble is too troublesome, so maybe this changes in the future.