Chapter 2: Installing Kafka

Question 1

By default, Kafka will automatically create a topic when a producer, consumer, or client attempts to interact with it. T/F

Answer 1

T

Question 2

A company is going to manage their topic creation using Pulumi. What Kafa configuration needs to be set so that producers, consumers, and clients cannot accidentally create topics without using Pulumi?

Answer 2

auto.create.topics.enable = false

Question 3

A majority of topics in a cluster will need a retention time of 2 hours. There are two ways of configuring Kafka to achieve this. What are they?

Answer 3

Set the retention time for each topic or set the default topic retention time

Question 4

A Kafka cluster has automatic topic creation enabled. How can the Kafka cluster be configured so that every topic will have at least 10 partitions?

Answer 4

Set the num.partitions parameter

Question 5

A Kafka cluster is using the default configuration parameters. A topic is created automatically. How many partitions will this topic have?

Answer 5

One

Question 6

It is possible to increase the number of partitions in an existing Kafka topic. T/F?

Answer 6

True

Question 7

It is possible to decrease the number of partitions in an existing Kafka topic. T/F?

Answer 7

False

Question 8

A Kafka cluster has the num.partitions parameter set to 10. How could I create a topic with only 5 partitions, without changing the num.partitions parameter?

Answer 8

Use manual topic creation

Question 9

How could I ensure that message load is evenly distributed among brokers?

Answer 9

Set num.partitions to the number of brokers

Question 10

How does the number of partitions affect throughput?

Answer 10

Throughput is maximized when the number of partitions is >= the number of brokers

Question 11

How many consumers can read from a particular partition at once ?

Answer 11

Zero or one

Question 12

Messages are being produced in the “foo” topic at a rate of 30 messages/second. A consumer of the “foo” topic can consume 10 messages per second. There are 5 of these consumers. How many partitions should the “foo” topic have in order to maximize throughput?

Answer 12

At least 3

Question 13

You are planning out a new topic. The messages in this topic will be partitioned by key. You predict that you will need at least 10 partitions for maximum throughput in the short-term, but future changes could require as much as 100 partitions for maximum throughput. How should you configure the topic? Why?

Answer 13

Set the number of partitions to 100. Increasing the number of partitions later out would be challenging.

Question 14

What are the negative consequences of using too many partitions?

Answer 14

Increased broker memory consumption. Increased time for leader elections.

Question 15

What is the runtime software required to run ZooKeeper and Kafka?

Answer 15

Java

Question 16

What are the ideal server counts in a ZooKeeper ensemble?

Answer 16

3, 5, or 7

Question 17

Why should ZooKeeper ensembles use an odd server count?

Answer 17

Because a majority of members must be working in order to respond to requests

Question 18

Why should a ZooKeeper ensemble not use a single server?

Answer 18

Because if that server went down, then the ensemble could not respond to requests

Question 19

Why should a ZooKeeper ensemble not use 2 servers?

Answer 19

Because if either server went down, then the ensemble would not respond to requests

Question 20

What are the risks of running a three-node ZooKeeper ensemble?

Answer 20

During maintenance, only two nodes will be available to respond, so if there was an unexpected outage then the ensemble would be unable to respond to requests

Question 21

What is the problem with running more than 7 nodes in a ZooKeeper ensemble?

Answer 21

The consensus protocol does not scale well to more than 7 nodes

Question 22

In a ZooKeeper configuration file, what does the initLimit setting do?

Answer 22

It sets the maximum time allowed for a follower to establish a connection with a leader

Question 23

A ZooKeeper configuration file contains the following:

tickTime=2000
initLimit=20

How many seconds will the follower be alloted to establish a connection with the leader?

Answer 23

40 seconds

Question 24

In a ZooKeeper configuration file, what does the syncLimit setting do?

Answer 24

It sets the maximum drift between followers and the leader

Question 25

How does a ZooKeeper server know its ID?

Answer 25

It reads the myid file in the data directory

Question 26

A ZooKeeper configuration file contains

server.1=zoo1.example.com:2888:3888

What does that do?

Answer 26

The ZooKeeper server with ID 1 has hostname zoo1.example.com, peer port 2888, and leader port 3888