Using confluent-kafka-go on MacOS M1

Hello,

TLDR;

brew install librdkafka openssl@3 pkg-config
export PKG_CONFIG_PATH="/opt/homebrew/opt/openssl@3/lib/pkgconfig"
go test -tags dynamic ./...

I’ve been transition from a Linux machine to a MacOS M1 machine at work and when I ran tests for a Golang project, I noticed that the test failed on modules depending on librdkafka.

Initially I’ve had problems with Kafka on MacOS M1 on Docker, since I was using an older image version that didn’t have any arm64 build, updating to images to version 7.2.1[1] fixed the issues.

This time however the problems was with my Golang dependencies and not the Docker containers. Running the tests resulted in:

go test ./...
[...]
ld: warning: ignoring file /Users/dnutiu/go/pkg/mod/github.com/confluentinc/confluent-kafka-go@v1.8.2/kafka/librdkafka_vendor/librdkafka_darwin.a, building for macOS-arm64 but attempting to link with file built for macOS-x86_64
Undefined symbols for architecture arm64:

Undefined symbols for architecture arm64” My guess is that the published confluent-kafka-go package does not contain (yet) symbols for arm64, to fix the issues you can use the module with another librdkafka.

When installing librdkafka formula from Homebrew the library is built for arm64 architecture. To install run:

brew install librdkafka openssl@3 pkg-config

Next, we’ll use a tool called pkg-config to tell librdkafka where to find the other libraries, since it depends on openssl we need to export PKG_CONFIG_PATH. To grab the value run:

brew info openssl
==> openssl@3: stable 3.0.5 (bottled) [keg-only]
Cryptography and SSL/TLS Toolkit
https://openssl.org/
/opt/homebrew/Cellar/openssl@3/3.0.5 (6,444 files, 27.9MB)
  Poured from bottle on 2022-08-31 at 14:10:49
From: https://github.com/Homebrew/homebrew-core/blob/HEAD/Formula/openssl@3.rb
License: Apache-2.0
==> Dependencies
Required: ca-certificates ✘
==> Caveats
A CA file has been bootstrapped using certificates from the system
keychain. To add additional certificates, place .pem files in
  /opt/homebrew/etc/openssl@3/certs

and run
  /opt/homebrew/opt/openssl@3/bin/c_rehash

openssl@3 is keg-only, which means it was not symlinked into /opt/homebrew,
because macOS provides LibreSSL.

If you need to have openssl@3 first in your PATH, run:
  echo 'export PATH="/opt/homebrew/opt/openssl@3/bin:$PATH"' >> ~/.zshrc

For compilers to find openssl@3 you may need to set:
  export LDFLAGS="-L/opt/homebrew/opt/openssl@3/lib"
  export CPPFLAGS="-I/opt/homebrew/opt/openssl@3/include"

For pkg-config to find openssl@3 you may need to set:
  export PKG_CONFIG_PATH="/opt/homebrew/opt/openssl@3/lib/pkgconfig"

Check that everything works by running:

pkg-config --libs --cflags rdkafka

-I/opt/homebrew/Cellar/openssl@3/3.0.5/include -I/opt/homebrew/Cellar/librdkafka/1.9.2/include -I/opt/homebrew/Cellar/zstd/1.5.2/include -I/opt/homebrew/Cellar/lz4/1.9.4/include -L/opt/homebrew/Cellar/librdkafka/1.9.2/lib -lrdkafka

Now, all you need to do is run your tests with the -tags dynamic flag, it will instruct confluent-kafka-go to use the librdkafka library that we’ve built from source.

Thanks for reading and happy hacking!🫶

Object Pool Pattern

Hi 👋

In this article we’ll talk about the Object Pool pattern in Golang.

The Object Pool pattern is a design pattern used in situations when constructing objects is a costly operation, for example building an HTTPClient or DatabaseClient object can take some time.

By having a pool of resources, the resources are requested from the pool when needed and then returned when not needed so they can be reused later.

Programs can benefit from this pattern because once the object is constructed when you need it again, you’ll just grab an instance instead of constructing it again from scratch.

In Golang this pattern is easily implemented with sync.Pool. Given a struct Resource struct, to implement an object pool we’ll need to pass the NewResource function to the pool.

To track how many active instances, we have of the object Resource, we use the counter variable.

Resource

var logger = log.Default()
var counter = 0
 
type Resource struct {
    id string
}
 
func NewResource() *Resource {
    logger.Printf("NewResource called")
    counter += 1
    return &Resource{id: fmt.Sprintf("Resource-%d", counter)}
}
 
func (r *Resource) doWork() {
    logger.Printf("%s doing work", r.id)
}
 

Let’s demo sync.Pool!

Demo 1️⃣

In the first demo, we get the resource from the pool, do some work and then put it back. By doing this one step at the time in the end we’ll end with just one Resource instance.

func demo1() {
	println("demo1")
	theResourcePool := sync.Pool{New: func() any {
		return NewResource()
	}}

	for i := 0; i < 10; i++ {
		item := theResourcePool.Get().(*Resource)
		item.doWork()
		theResourcePool.Put(item)
	}

	println("done", counter)
}

Output

demo1
2022/08/17 22:38:59 NewResource called
2022/08/17 22:38:59 Resource-1 doing work
2022/08/17 22:38:59 Resource-1 doing work
2022/08/17 22:38:59 Resource-1 doing work
2022/08/17 22:38:59 Resource-1 doing work
2022/08/17 22:38:59 Resource-1 doing work
2022/08/17 22:38:59 Resource-1 doing work
2022/08/17 22:38:59 Resource-1 doing work
2022/08/17 22:38:59 Resource-1 doing work
2022/08/17 22:38:59 Resource-1 doing work
2022/08/17 22:38:59 Resource-1 doing work
done 1

Resource-1 is the only instance that does work.

Demo 2️⃣

In demo2 we spawn 10 goroutines, that use the pool. Since all goroutines start roughly at the same time and require a resource to doWork, in the end the pool will have 10 Resource instances.

func demo2() {
	println("demo2")
	wg := sync.WaitGroup{}
	theResourcePool := sync.Pool{New: func() any {
		return NewResource()
	}}

	for i := 0; i < 10; i++ {
		wg.Add(1)
		go func() {
			defer wg.Done()
			item := theResourcePool.Get().(*Resource)
			item.doWork()
			theResourcePool.Put(item)
		}()

	}
	wg.Wait()

	println("done", counter)
}

Output

demo2
2022/08/17 22:41:12 NewResource called
2022/08/17 22:41:12 NewResource called
2022/08/17 22:41:12 NewResource called
2022/08/17 22:41:12 Resource-3 doing work
2022/08/17 22:41:12 NewResource called
2022/08/17 22:41:12 Resource-4 doing work
2022/08/17 22:41:12 NewResource called
2022/08/17 22:41:12 Resource-5 doing work
2022/08/17 22:41:12 NewResource called
2022/08/17 22:41:12 Resource-6 doing work
2022/08/17 22:41:12 NewResource called
2022/08/17 22:41:12 Resource-7 doing work
2022/08/17 22:41:12 NewResource called
2022/08/17 22:41:12 Resource-8 doing work
2022/08/17 22:41:12 NewResource called
2022/08/17 22:41:12 NewResource called
2022/08/17 22:41:12 Resource-1 doing work
2022/08/17 22:41:12 Resource-2 doing work
2022/08/17 22:41:12 Resource-9 doing work
2022/08/17 22:41:12 Resource-10 doing work
done 10

Demo 3️⃣

In demo3 doing the same thing we did in demo2 with some random sleeps in between, some goroutines are faster and others are slower. The faster goroutines will also return the resource faster to the pool and slower goroutines which start at a later time will reuse the resource instead of creating a new one.

func demo3() {
	println("demo2")
	wg := sync.WaitGroup{}
	theResourcePool := sync.Pool{New: func() any {
		return NewResource()
	}}

	for i := 0; i < 10; i++ {
		wg.Add(1)
		go func() {
			defer wg.Done()
			time.Sleep(time.Duration(rand.Intn(900)+100) * time.Millisecond)
			item := theResourcePool.Get().(*Resource)
			item.doWork()
			time.Sleep(time.Duration(rand.Intn(100)+100) * time.Millisecond)
			theResourcePool.Put(item)
		}()

	}
	wg.Wait()

	println("done", counter)
}

Output

demo2
2022/08/17 22:42:35 NewResource called
2022/08/17 22:42:35 Resource-1 doing work
2022/08/17 22:42:35 NewResource called
2022/08/17 22:42:35 Resource-2 doing work
2022/08/17 22:42:35 NewResource called
2022/08/17 22:42:35 Resource-3 doing work
2022/08/17 22:42:36 Resource-1 doing work
2022/08/17 22:42:36 Resource-2 doing work
2022/08/17 22:42:36 Resource-3 doing work
2022/08/17 22:42:36 Resource-1 doing work
2022/08/17 22:42:36 NewResource called
2022/08/17 22:42:36 Resource-4 doing work
2022/08/17 22:42:36 NewResource called
2022/08/17 22:42:36 Resource-5 doing work
2022/08/17 22:42:36 Resource-2 doing work
done 5

Only 5 Resource instances have been created at this time.

Conclusion

The object pool pattern is a great pattern when you need to reuse an instance of an object. Constructing the object every time can be slow.

In Go we have sync.pool which implements the Object Pool pattern for us, we just need to give it a New function that returns a pointer.

Thanks for reading! 📚

References

Full Code

package main

import (
	"fmt"
	"log"
	"math/rand"
	"sync"
	"time"
)

var logger = log.Default()
var counter = 0

type Resource struct {
	id string
}

func NewResource() *Resource {
	logger.Printf("NewResource called")
	counter += 1
	return &Resource{id: fmt.Sprintf("Resource-%d", counter)}
}

func (r *Resource) doWork() {
	logger.Printf("%s doing work", r.id)
}

func demo1() {
	println("demo1")
	theResourcePool := sync.Pool{New: func() any {
		return NewResource()
	}}

	for i := 0; i < 10; i++ {
		item := theResourcePool.Get().(*Resource)
		item.doWork()
		theResourcePool.Put(item)
	}

	println("done", counter)
}

func demo2() {
	println("demo2")
	wg := sync.WaitGroup{}
	theResourcePool := sync.Pool{New: func() any {
		return NewResource()
	}}

	for i := 0; i < 10; i++ {
		wg.Add(1)
		go func() {
			defer wg.Done()
			item := theResourcePool.Get().(*Resource)
			item.doWork()
			theResourcePool.Put(item)
		}()

	}
	wg.Wait()

	println("done", counter)
}

func demo3() {
	println("demo2")
	wg := sync.WaitGroup{}
	theResourcePool := sync.Pool{New: func() any {
		return NewResource()
	}}

	for i := 0; i < 10; i++ {
		wg.Add(1)
		go func() {
			defer wg.Done()
			time.Sleep(time.Duration(rand.Intn(900)+100) * time.Millisecond)
			item := theResourcePool.Get().(*Resource)
			item.doWork()
			time.Sleep(time.Duration(rand.Intn(100)+100) * time.Millisecond)
			theResourcePool.Put(item)
		}()

	}
	wg.Wait()

	println("done", counter)
}

func main() {
	demo1()
	//demo2()
	//demo3()
}

Testing Tips: Avoid sleep in tests

Hi ðŸ‘‹,

In this article I wanna show a testing tip that I’ve recently learned myself by reading Software Engineering at Google: Lessons Learned from Programming Over Time. The technique improved the way I write unit tests.

When I’m writing bigger unit tests, I have execute something in the background, like for example publishing a message to a message broker, wait for the message to be published and then consume it to test that what I published is correct.

When waiting for the message to be published or any other operation that required waiting in tests I used to call a sleep function, for a second or two, this is decent for few tests but if your tests grow then this approach does not scale well. Imagine if you’re having 50 tests and each test sleeps for one second, it would take at least 50 seconds to run the test suite, which is a lot of wasted time.

The better approach is to use a timeout and polling, you can poll at every millisecond to see if your test has done what you wanted to do instead of sleeping, this will improve the tests and reduce the execution time by a lot!

Let’s demonstrate this will a small example using the Golang programming language, I’m not going to use any external dependencies to demonstrate this technique but you can apply it everywhere you’re calling something that blocks or if you need to wait for something.

What we’re going to test is a simple struct with a method that blocks and modifies a field.

import (
	"math/rand"
	"time"
)

type SystemUnderTest struct {
	Result string
}

func (s *SystemUnderTest) SetResult() {
	go func() {
		time.Sleep(time.Duration(rand.Intn(3000)) * time.Millisecond)
		s.Result = "the_result"
	}()
}

func (s *SystemUnderTest) GetData() string {
	time.Sleep(time.Duration(rand.Intn(3000)) * time.Millisecond)
	return "the_data"
}

This is the not ideal way of testing it:

// A not very ideal way to test SetResult
func Test_SystemUnderTest_SetResult_NotIdeal(t *testing.T) {
	sut := SystemUnderTest{}
	sut.SetResult()

	time.Sleep(4 * time.Second)

	if sut.Result != "the_result" {
		t.Fatalf("Result not equal, want %s got %s", "the_result", sut.Result)
	}
}

SetResults takes between 0 to 3 seconds to run, since we’re waiting for the result we’re sleeping for 4 seconds.

=== RUN   Test_SystemUnderTest_SetResult_NotIdeal
--- PASS: Test_SystemUnderTest_SetResult_NotIdeal (4.00s)
PASS

A better way is to write a simple loop and poll for the result:

// A better way of testing the code
func Test_SystemUnderTest_SetResult(t *testing.T) {
	sut := SystemUnderTest{}
	sut.SetResult()

	passedMilliseconds := 0
	for {
		if passedMilliseconds > 4000 {
			t.Fatalf("timeout reached")
		}
		passedMilliseconds += 1
		time.Sleep(1 * time.Millisecond)
		if sut.Result != "" {
			break
		}
	}
	if sut.Result != "the_result" {
		t.Fatalf("Result not equal, want %s got %s", "the_result", sut.Result)
	}
}

Writing a loop and polling for the result will make the test more complex but it will execute faster. In this case the benefits outweigh the downsides.

=== RUN   Test_SystemUnderTest_SetResult
--- PASS: Test_SystemUnderTest_SetResult (2.08s)
PASS

If the language permits we can also use channels, let’s change the following function that returns a result after a random amount of time and test it.

func Test_SystemUnderTest_GetData(t *testing.T) {
	sut := SystemUnderTest{}

	timeoutTicker := time.NewTicker(5 * time.Second)
	result := make(chan string)

	// Get result when ready
	go func() {
		result <- sut.GetData()
	}()

	select {
	case <-timeoutTicker.C:
		t.Fatal("timeout reached")
	case actual := <-result:
		if actual != "the_data" {
			t.Fatalf("Data not equal, want: %s, got %s", "the_data", actual)
		}
	}
}

We avoided writing a loop with the use of a ticker and select.

In another case you may need to test HTTP calls on the local machine or any other library. Look for timeout options.

Go’s HTTP library let’s you specify a custom timeout for every call you make:

	client := http.Client{
		Timeout: 50 * time.Millisecond,
	}
	response, err := client.Get("http://localhost:9999/metrics")
	...

In Conclusion

Avoid the use of sleep in tests, try polling for the result instead or check if the blocking functions have parameters or can be configured to stop the execution after a timeout period.

Thanks for reading and I hope you’ve enjoyed this article! 🍻