Why Acknowledgment modes (acks=0, 1, all) in Kafka? - Purpose & Use Cases

The Big Idea

What if your important messages vanish without you knowing? Discover how acknowledgment modes keep them safe!

The Scenario

Imagine you are sending important messages to friends by mail, but you never get a confirmation if they received it. Sometimes the mail gets lost, and you don't know if your message arrived.

The Problem

Without a way to confirm delivery, you might resend messages multiple times or miss that some never arrived. This causes confusion, wasted effort, and lost information.

The Solution

Kafka's acknowledgment modes let you choose how much confirmation you want when sending messages. This way, you can balance speed and safety, knowing when your messages are safely received.

Before vs After

✗ Before

producer.send(message)
// no confirmation, might lose messages

✓ After

producer.send(message)
// producer configured with acks='all', waits for all replicas to confirm, safer delivery

What It Enables

You can control message delivery guarantees, making your system reliable without slowing it down unnecessarily.

Real Life Example

In a banking app, you want to be sure transaction messages are fully saved before moving on, so you use acks='all' to avoid losing money transfer records.

Key Takeaways

Manual sending risks losing messages without notice.

Acknowledgment modes provide delivery confirmation levels.

Choosing the right acks balances speed and reliability.

Practice

1. Which acks mode in Kafka provides the fastest message sending but with the least safety?

easy

A. acks=0

B. acks=1

C. acks=all

D. None of the above

Solution

1. Step 1: Understand acks=0 behavior

   This mode means the producer does not wait for any acknowledgment from the broker, so it sends messages fastest but with no guarantee of delivery.

2. Step 2: Compare with other modes

   acks=1 waits for leader confirmation, and acks=all waits for all replicas, both slower but safer.

3. Final Answer:

   acks=0 -> Option A

4. Quick Check:

   Fastest but least safe = acks=0 [OK]

Hint: No wait for confirmation means fastest mode is acks=0 [OK]

Common Mistakes:

• Confusing acks=0 with acks=1
• Thinking acks=all is fastest
• Assuming acks=0 waits for leader

2. Which of the following is the correct way to set the acknowledgment mode to wait for all replicas in Kafka producer configuration?

easy

A. "acks=1"

B. "acks=0"

C. "acks=all"

D. "ack=all"

Solution

1. Step 1: Recall correct property name and value

   The property is acks (plural), and to wait for all replicas, the value is all.

2. Step 2: Check options for syntax correctness

   "acks=all" uses correct key and value: "acks=all". "ack=all" has a typo "ack" instead of "acks".

3. Final Answer:

   "acks=all" -> Option C

4. Quick Check:

   Wait for all replicas = acks=all [OK]

Hint: Remember plural 'acks' and value 'all' for all replicas [OK]

Common Mistakes:

• Using 'ack' instead of 'acks'
• Confusing 'acks=1' with 'acks=all'
• Setting 'acks=0' for all replicas

3. Consider this Kafka producer code snippet:

props.put("acks", "1");
producer.send(record);
System.out.println("Message sent");

What does acks=1 guarantee before printing "Message sent"?

medium

A. Message is stored only in producer buffer

B. No guarantee; message may be lost

C. All replicas have confirmed the message

D. Leader broker has confirmed the message

Solution

1. Step 1: Understand acks=1 meaning

   This mode waits for the leader broker to confirm the message is received and written to its log.

2. Step 2: Compare with other options

   acks=0 gives no guarantee, acks=all waits for all replicas, and producer buffer storage is unrelated.

3. Final Answer:

   Leader broker has confirmed the message -> Option D

4. Quick Check:

   acks=1 means leader confirmation [OK]

Hint: acks=1 waits for leader confirmation only [OK]

Common Mistakes:

• Thinking acks=1 waits for all replicas
• Assuming no guarantee with acks=1
• Confusing producer buffer with broker confirmation

4. You have this Kafka producer configuration:

props.put("acks", "alll");

What is the problem with this configuration?

medium

A. Typo in the value; should be "all"

B. Typo in the key; should be "ack"

C. No problem; "alll" is valid

D. Should be an integer, not a string

Solution

1. Step 1: Check the value spelling

   The correct value for waiting all replicas is "all", but here it is misspelled as "alll" with an extra 'l'.

2. Step 2: Verify key and value types

   The key "acks" is correct and the value should be a string. So the problem is only the typo in the value.

3. Final Answer:

   Typo in the value; should be "all" -> Option A

4. Quick Check:

   Value must be "all", not "alll" [OK]

Hint: Check spelling of 'all' carefully in acks value [OK]

Common Mistakes:

• Assuming 'alll' is valid
• Changing key to 'ack' instead of 'acks'
• Using integer instead of string for acks

5. You want to ensure maximum data safety in Kafka producer but can tolerate some delay. Which acks setting should you use and why?

hard

A. acks=1 because leader confirmation is enough

B. acks=all because it waits for all replicas

C. acks=0 because it is fastest

D. acks=-1 because it disables acknowledgments

Solution

1. Step 1: Identify safety vs speed tradeoff

   acks=all waits for all replicas to confirm, ensuring maximum data safety but slower performance.

2. Step 2: Evaluate other options

   acks=0 is fastest but unsafe, acks=1 waits only for leader, and acks=-1 is invalid or equivalent to acks=all but less clear.

3. Final Answer:

   acks=all because it waits for all replicas -> Option B

4. Quick Check:

   Max safety = acks=all [OK]

Hint: Max safety means waiting for all replicas: acks=all [OK]

Common Mistakes:

• Choosing acks=0 for safety
• Confusing acks=-1 as disabling acknowledgments
• Thinking leader confirmation is max safety