I saw that Cassandra client needs an array of hosts.

For example, Python uses this:

from cassandra.cluster import Cluster

cluster = Cluster(['192.168.0.1', '192.168.0.2'])

• source: http://datastax.github.io/python-driver/getting_started.html

Question 1: Why do I need to pass these nodes?

Question 2: Do I need to pass all nodes? Or is one sufficient? (All nodes have the information about all other nodes, right?)

Question 3: Does the client choose the best node to connect knowing all nodes? Does the client know what data is stored in each node?

Question 4: I'm starting to use Cassandra for the first time, and I'm using Kubernetes for the first time. I deployed a Cassandra cluster with 3 Cassandra nodes. I deployed another one machine and in this machine, I want to connect to Cassandra by a Python Cassandra client. Do I need to pass all the Cassandra IPs to Python Cassandra client? Or is it sufficient to put the Cassandra DNS given by Kubernetes?

For example, when I run a dig command, I know all the Cassandra IPs. I don't know if it's sufficient to pass this DNS to the client

# dig cassandra.default.svc.cluster.local

The IPs are 10.32.1.19, 10.32.1.24, 10.32.2.24

; <<>> DiG 9.10.3-P4-Debian <<>> cassandra.default.svc.cluster.local
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 18340
;; flags: qr aa rd ra; QUERY: 1, ANSWER: 3, AUTHORITY: 0, ADDITIONAL: 0

;; QUESTION SECTION:
;cassandra.default.svc.cluster.local. IN    A

;; ANSWER SECTION:
cassandra.default.svc.cluster.local. 30 IN A 10.32.1.19
cassandra.default.svc.cluster.local. 30 IN A 10.32.1.24
cassandra.default.svc.cluster.local. 30 IN A 10.32.2.24

;; Query time: 2 msec
;; SERVER: 10.35.240.10#53(10.35.240.10)
;; WHEN: Thu Apr 04 16:08:06 UTC 2019
;; MSG SIZE  rcvd: 125

What are the disadvantages of using for example:

from cassandra.cluster import Cluster

cluster = Cluster(['cassandra.default.svc.cluster.local'])