Extend Cosmos performance benchmark to analyse memory usage

[tatiana]

In Cosmos, we have benchmark scripts that test task throughput using synthetic data and using Airflow local executor:
https://github.com/astronomer/astronomer-cosmos/blob/main/.github/workflows/test.yml#L350
https://github.com/astronomer/astronomer-cosmos/blob/main/tests/perf/test_performance.py

While those are very helpful, we currently don't analyse memory consumption.

It would be great if, from time to time, we could run a performance benchmark in Astro and we could identify the memory footprint per task,

[tatiana]

Related task: #1471

[josix]

Would memray/pytest-memary suit our needs? Perhaps we could generate a memory footprint report alongside the performance test—it might help identify where the overhead occurs. If so, I'd be happy to help create a POC for it.

[tatiana]

Yes, @josix , I think that would be a great start!

An approach that @sbaldassin used in the past was to dynamically create N tasks and subclass Cosmos Operators so they would use psutil and somehow log the mem_info. This example is a function used by PythonOperator, but we could try to use a similar approach in Cosmos:

(credit of this code to @sbaldassin:

def write_temp_files():
    process = psutil.Process()
    temp_dir = tempfile.mkdtemp()
    print(f"Temporary directory created: {temp_dir}")

    try:
        for i in range(10):  # Create 10 temporary files
            file_path = os.path.join(temp_dir, f"temp_file_{i}.txt")
            with open(file_path, "w") as temp_file:
                temp_file.write("X" * 10**6)  # Write 1 MB of data

            mem_info = process.memory_info()
            print(f"Step {i + 1}: Memory Usage: {mem_info.rss / (1024 * 1024):.2f} MB")

            time.sleep(1)  # Simulate delay between file writes

    finally:
        for file_name in os.listdir(temp_dir):
            os.remove(os.path.join(temp_dir, file_name))
        os.rmdir(temp_dir)
        print(f"Temporary directory deleted: {temp_dir}")

[github-actions]

This issue is stale because it has been open for 30 days with no activity.