Monitor AWS ELB (Classic) with SigNoz

Log Collection

This logs setup uses an AWS Lambda function to forward ELB logs from S3 to SigNoz.

Creating / Configuring Your S3 Bucket

To accomplish the task described, please follow these steps:

1. Creating an S3 Bucket:
   • Sign in to the AWS Management Console.
   • Navigate to the Amazon S3 service.
   • Click on Create bucket.
   • Enter a unique bucket name, select the region, and configure any additional settings if needed (such as versioning, logging, etc.).
   • Click Create bucket to finalize the creation process.

2. Uploading Data to the S3 Bucket:

   • After creating the bucket, navigate to the S3 Management Console.
   • Click on the Upload button.
   • Select the files you wish to upload from your local machine.
   • Optionally, configure settings like encryption, permissions, etc.
   • Click Upload to upload the selected files to the bucket.

3. File Formats:

   • You can upload files in various formats such as .json, .csv, .log, .gz, .zip, etc.
   • If you have configured Elastic Load Balancer (ELB) logging, ELB logs are automatically saved in .gzip format in your S3 bucket.

To move forward, we assume that you already have some data in your S3 bucket.

The general header(table) format of ELB logs is:

elb_headers= ["type","time","elb","client_port","target_port",
"request_processing_time","target_processing_time","response_processing_time",
"elb_status_code","target_status_code","received_bytes","sent_bytes","request",
"user_agent","ssl_cipher","ssl_protocol","target_group_arn","trace_idd",
"domain_name","chosen_cert_arn","matched_rule_priority",
"request_creation_time","actions_executed","redirect_url",
"error_reason","target_port_list","target_status_code_list",
"classification","classification_reason"]

[
	{
		"timestamp": <uint64>,
		"trace_id": <hex string>,
		"span_id": <hex string>,
		"trace_flags": <int>
		"severity_text": <string>,
		"severity_number": <int>,
		"attributes": <map>,
		"resources": <map>,
		"body": <string>,
	}
]

For more details, see Exporting logs to SigNoz using HTTP.

trace_id here is of the format <hex string> and because we are sending everything as string json for simplicity and less data processing and parsing, hence the change in field name from trace_id to trace_idd. If you use trace_id and send the trace_id as a string, you’ll get 400 error, this can be resolved by further logs formatting

Understanding How Lambda Function Works

When you successfully attach your lambda function with the S3 bucket and configure it correctly, any new addition / deletion / copy / PUT etc, requests made to the S3 bucket will trigger the lambda function and the code written in the lambda function will get executed.

Creating a Lambda Function

Follow these steps to create the lambda function:

Step 1: Go to your AWS console and search for AWS Lambda, go to Functions and click on Create Function.

Step 2: Choose the Author from scratch checkbox and proceed to fill in the function name.

Step 3: Choose Python 3.x as the Runtime version, x86_64 as Architecture (preferably), and keep other settings as default. Select Create a new role with basic Lambda permissionsfor now, we’ll requiring more permissions down the lane. So for now, select this option.

Step 4: Once you are done configuring the lambda function, your Lambda function is created.

Configuring Policies for Lambda Function

As said in Step 4 previously, we need extra permissions in order to access the S3 Bucket for execution of our Lambda code, follow along to set it up.

Step 1: Scroll down from your Lambda page, you’ll see a few tabs there. Go to Configurations and select Permissions from the left sidebar.

Step 2: Click on the Execution Role name link just under Role name, it will take us to AWS IAM page. Here we will add policies to get full S3 access. Once here, click on the Add permissions button and select Attach policies from the drop down list.

Step 3: Search “S3” and you’ll a policy named AmazonS3FullAccessselect that and proceed.

Please refer to the image above as a comprehensive guide to the policy names that you may consider adding to your Lambda function. Failure to include these policies could result in insufficient privileges, potentially hindering the function's ability to perform necessary operations within the AWS environment.

Congrats, you are just done with one of the major hurdle in running your code. Now, let’s add a trigger.

Adding Triggers

You need to use the Lambda console to build a trigger so that your function can be called immediately by another AWS service (S3, in our case). A trigger is a resource you set up to enable your function to be called by another AWS service upon the occurrence of specific events or conditions.

A function may have more than one trigger. Every trigger functions as a client, independently calling your method, and Lambda transfers data from a single trigger to each event it passes to your function.

To setup the trigger, follow these steps:

Step 1: Click on the + Add trigger button from the Lambda console.

Step 2: Select S3 from the first drop down of AWS services list. Pick your S3 bucket for the second field.

Step 3: For the Event types field, you can select any number of options you wish. The trigger will occur depending upon what option(s) you choose here. By default, the All object create events will be selected.

Verify the settings and click on Add button at bottom right to add this trigger.

Adding Request Layer

We will be using Python's requests module which is not included by default in Lambda.

To utilize Python's requests module within a Lambda function, it's necessary to explicitly add it as a layer. While there may be alternative approaches, it's advisable to adhere to established practices that have been thoroughly tested and proven effective. Therefore, we will proceed with adding requests as a layer to ensure reliable functionality within the Lambda environment.

Step 1: Follow the steps below to create a zip of the request module and add it as a layer to make it work on AWS lambda.

The commands you’d need:

# make a new directory
mkdir python
# move into that directory
cd python

# install requests module
pip install --target . requests
# zip the contents under the name dependencies.zip
zip -r dependencies.zip ../python 

Step 2: To upload your zip file, go to AWS Lambda > Layers and click on Create Layer. [Not inside your specific Lambda function, just the landing page of AWS Lambda].

Step 3: You’ll be redirected to Layer configurations page, here, give a name to your layer, an optional description, select Upload a .zip file , click on Upload and locate the requirements.zip file.

Step 4: Select your desired architecture and pick Python 3.x as your runtime. Hit Create. Your layer has now been created. Now lets connect it to our Lambda function which we created to send logs to SigNoz.

Step 5: Go to your Lambda function, scroll down to Layers section and on the right of it, you’ll find a button that says Add a layer to click on.

Step 6: Pick Custom layers from the checkbox and select your custom layer from the given drop down below and then click on the button Add.

Congratulations, the requests module has been successfully integrated into your code area. By adding this layer, you have resolved the 'request module not found error' that would have otherwise occurred.

The Lambda Function

Now, we come to the pivotal section of this document: the code implementation.

The Python script's primary function revolves around retrieving gzipped log files stored within an Amazon S3 bucket. Subsequently, it decompresses these files, transforms individual log entries into JSON objects, and transmits the resultant JSON data to a predetermined HTTP endpoint.

Below is the comprehensive code along with detailed comments for clarity:

import json
import gzip
import boto3 
import requests
import shlex

# Create an S3 client
s3 = boto3.client('s3')

# Function to convert a log line into a JSON object
def convert_log_line_to_json(line):
    # Define the headers to be used for the JSON keys
    headers = ["type", "time", "elb", "client:port", "target:port", "request_processing_time", "target_processing_time", "response_processing_time", "elb_status_code", "target_status_code", "received_bytes", "sent_bytes", "request","user_agent", "ssl_cipher", "ssl_protocol", "target_group_arn", "trace_idd", "domain_name", "chosen_cert_arn", "matched_rule_priority", "request_creation_time", "actions_executed", "redirect_url", "error_reason", "target:port_list", "target_status_code_list", "classification", "classification_reason"]

    # Split the log line using shell-like syntax (keeping quotes, etc.)
    res = shlex.split(line, posix=False)

    # Create a dictionary by zipping headers and log line parts
    return dict(zip(headers, res))

# Lambda function handler
def lambda_handler(event, context):
    # S3 bucket name
    bucket_name = '<name_of_your_bucket>'

    # List all objects in the specified S3 bucket
    response = s3.list_objects_v2(Bucket=bucket_name)

    # Iterate through each object in the bucket
    for obj in response['Contents']:
        # Check if the object is a gzipped log file
        if obj['Key'].endswith('.log.gz'):
            file_key = obj['Key']

            # Download the gzipped file content
            file_obj = s3.get_object(Bucket=bucket_name, Key=file_key)
            file_content = file_obj['Body'].read()

            # Decompress the gzipped content
            decompressed_content = gzip.decompress(file_content)

            # Convert bytes to string
            json_data = str(decompressed_content, encoding='utf-8')

            # Split the string into lines
            lines = json_data.strip().split('\n')

            # Convert the list of strings into a JSON-formatted string
            result = json.dumps(lines, indent=2)

            # Load the JSON-formatted string into a list of strings
            list_of_strings = json.loads(result)

            # Convert each log line string into a JSON object
            json_data = [convert_log_line_to_json(line) for line in list_of_strings]
            
            req_headers = {
                     'signoz-ingestion-key': '<your-ingestion-key>',
                     'Content-Type': 'application/json'
                }
            # Specify the HTTP endpoint for sending the data
            http_url = 'https://ingest.<region>.signoz.cloud:443/logs/json'  # Replace `<region>` with your [SigNoz Cloud region](https://signoz.io/docs/ingestion/signoz-cloud/overview/#endpoint) 

            # Send the JSON data to the specified HTTP endpoint
            response = requests.post(http_url, json=json_data, headers=req_headers)

            # Print information about the sent data and the response received
            print(f"Sent data to {http_url}. Response: {response.status_code}")

Verify these values:

• <name_of_your_bucket>: The name of your S3 bucket containing ELB logs.
• <your-ingestion-key>: Your SigNoz ingestion key.
• <region>: Your SigNoz Cloud region

Here’s how a raw, unprocessed ELB log line looks like:

https 2024-01-01T23:58:03.391277Z app/abc-prod-alb/0b46e552ds5b44da 35.244.22.76:41802 192.1.0.114:80 0.000 1.077 0.000 200 200 1430 923 "POST https://api.abcs.com:4463/suporodv2/v1/get-result/ HTTP/1.1" "SFDC-Callout/59.0" ECDHE-RSA-AES128-GCM-SHA256 TLSv1.2 arn:aws:elasticloadbalancing:ap-south-1:8429181216651:targetgroup/ecs-Sabc-P-Private-SU-API/02f1623fsddec2691ce "Root=1-65343518a-72123e913f71cb2e20213a3ea9" "api.example-sabcs.com" "session-reused" 98 2024-01-01T23:58:02.313000Z "forward" "-" "-" "192.1.1.114:80” "200" "-" "-"

In the code, each field corresponds to a header. The purpose of the code above is to transmit Elastic Load Balancer (ELB) logs to the SigNoz endpoint.

Other than the above explanation and the code comments, in a nutshell, what this code does is:

Sends the parsable content of ENTIRE S3 bucket whenever the lambda function gets triggered. It gets triggered by the condition you set above. Let’s mention that again here.

Step 3: For the Event types field, you can select any number of options you wish. The trigger will occur depending upon what option(s) you choose here. By default, the All object create events will be selected.

Lets say you add something to your S3 bucket, it may / may not trigger this lambda function or if you have setup your s3 as if it automatically stores all your ELB/VPC logs, segregated in different folders, so whenever any new log gets added, the function will get triggered and send all the S3 data.

This is obviously not what everyone expects, ideal case would be to have a mass log transfer once the first connection is made to SigNoz otel-collector (which then they later get stored in gp2/gp3 storageClass of EBS), and then send logs lines of only the recently logged one.

To achieve this functionality, you need to add few conditions to the code.

1. Assuming all standard log lines have a timestamp field.
2. Parse and select the timestamp field from the log line and add it before the response = requests.post(http_url, json=json_data) line as a if else condition to only send logs which are x days older (say 3 days).

So, the function now will first check the log timestamp and only send those logs which are 3 days older (say) or even a few hours old.

Let’s consider the below pseudo code for better understanding:

   from datetime import datetime, timedelta

   # Your given timestamp
   given_timestamp_str = "2024-01-01T23:58:02.231919Z"
   given_timestamp = datetime.fromisoformat(given_timestamp_str.replace('Z', '+00:00'))

   # Current time
   current_time = datetime.utcnow()

   # Calculate the time difference
   time_difference = current_time - given_timestamp

   # Check if the time difference is less than 3 days
   if time_difference < timedelta(days=3):
       # Run your specific function here
       print("Running the specific function.")
   # ADD THE response = requests.post(http_url, json=json_data) LINE HERE

   else:
       print("Time difference exceeds 3 days. Function will not run.")

Feel free to modify any part of the code according to your requirements.

Running the Code Locally

If you want to run the entire setup locally in your laptop for testing purposes. Here’s the reference code for you:

import os
import gzip
import json
import requests
import shlex

def convert_log_line_to_json(line):
    headers= ["type","time","elb","client_port","target_port","request_processing_time","target_processing_time","response_processing_time","elb_status_code","target_status_code","received_bytes","sent_bytes","request","user_agent","ssl_cipher","ssl_protocol","target_group_arn","trace_idd","domain_name","chosen_cert_arn","matched_rule_priority","request_creation_time","actions_executed","redirect_url","error_reason","target_port_list","target_status_code_list","classification","classification_reason"]
    res = shlex.split(line, posix = False)
    return dict(zip(headers, res))

def process_log_file(file_path):
    with gzip.open(file_path, 'r') as f:
        log_data = f.read().decode('utf-8') 
        lines = log_data.strip().split('\n')
        result = json.dumps(lines, indent=2)
        list_of_strings = json.loads(result)
        json_data = [convert_log_line_to_json(line) for line in list_of_strings]
        req_headers = {
                     'signoz-ingestion-key': '<your-ingestion-key>',
                     'Content-Type': 'application/json'
                }
        http_url = 'https://ingest.<region>.signoz.cloud:443/logs/json'
        response = requests.post(http_url, json=json_data, headers=req_headers)

def main():
    root_folder = '<folder_name>'

    for root, _, files in os.walk(root_folder):
        for file in files:
            if file.endswith('.log.gz'):
                file_path = os.path.join(root, file)
                process_log_file(file_path)

if __name__ == '__main__':
    main()

Testing Your Lambda Function

Once you've finished writing your code, it's crucial to deploy it and conduct thorough testing to ensure its functionality. Before proceeding with deployment and testing, it's important to consider adjusting the timeout setting for your Lambda function. This adjustment is necessary because the process of transferring data from S3 to an external endpoint may take several minutes, exceeding the default Lambda timeout of 3 seconds.

To extend the timeout duration, follow these steps:

1. Navigate to the Lambda function configuration.
2. Access the "General Configuration" section.
3. Click on the "Edit" button to modify settings.
4. Increase the timeout value to a duration exceeding 10 minutes. Typically, the code execution completes within 1-4 minutes at most.

By adjusting the timeout setting, you ensure that your Lambda function has sufficient time to complete the data transfer process without encountering timeouts. This proactive measure enhances the reliability and effectiveness of your deployed solution.

Once you've finished adjusting the timeout setting, navigate to the code editor for your Lambda function. Locate the 'test' button, and from the dropdown menu, select the option labeled 'Configure test events.' Create a new test case by specifying it as an S3 PUT event, then save your configuration.

You're now prepared to proceed. Whenever you make alterations to the code and wish to evaluate them, follow these steps: Deploy the code first (equivalent to pressing 'Save'), and once it's fully deployed, proceed to click on the 'Test' button.

Below is an image showing the process of transmitting VPC logs (excluding ELB logs) to the SigNoz endpoint.

Test Case and Output

If the logs are sent successfully, here's how they'll be transmitted. The following output displays the JSON-formatted data as we've printed it to visualize the sent information.

Validate

To verify that your ELB logs are being sent correctly to SigNoz:

1. Navigate to Logs > Logs Explorer in SigNoz.
2. You should see your ELB logs appearing in the logs list.
3. Click on any log line to view its detailed information and verify the fields.

Metrics Collection

To collect ELB metrics manually, you can use the Prometheus CloudWatch Exporter.

Prerequisites

Before proceeding, ensure you have:

• OpenTelemetry Collector installed and configured. See Get Started with OTel Collector.
• Java 11 or higher installed on the host machine (for JAR-based setup), or Docker (for container-based setup).
• AWS credentials configured via environment variables (AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY), IAM role, or ~/.aws/credentials.
• IAM permissions for the credentials:
  • cloudwatch:ListMetrics
  • cloudwatch:GetMetricStatistics

Where to Run the CloudWatch Exporter

The CloudWatch Exporter should run on a machine that:

1. Has network access to AWS CloudWatch APIs.
2. Has AWS credentials configured.
3. Is network-accessible from your OpenTelemetry Collector.

For SigNoz Cloud users, run the exporter on any EC2 instance, VM, or container with AWS credentials. The OTel Collector on the same host will forward metrics to SigNoz Cloud.

For Self-Hosted users, run the exporter on the same host as your OTel Collector, or ensure network connectivity between them.

Step 1: Create Configuration File

Create a file named elb-metrics.yaml with the following configuration:

region: <aws-region>
metrics:
  - aws_namespace: AWS/ELB
    aws_metric_name: RequestCount
    aws_dimensions: [LoadBalancerName]
    aws_statistics: [Sum]

  - aws_namespace: AWS/ELB
    aws_metric_name: Latency
    aws_dimensions: [LoadBalancerName]
    aws_statistics: [Average]
    aws_extended_statistics: [p99]

  - aws_namespace: AWS/ELB
    aws_metric_name: HealthyHostCount
    aws_dimensions: [LoadBalancerName]
    aws_statistics: [Average]

  - aws_namespace: AWS/ELB
    aws_metric_name: UnHealthyHostCount
    aws_dimensions: [LoadBalancerName]
    aws_statistics: [Average]

  - aws_namespace: AWS/ELB
    aws_metric_name: HTTPCode_Backend_2XX
    aws_dimensions: [LoadBalancerName]
    aws_statistics: [Sum]

  - aws_namespace: AWS/ELB
    aws_metric_name: HTTPCode_Backend_4XX
    aws_dimensions: [LoadBalancerName]
    aws_statistics: [Sum]

  - aws_namespace: AWS/ELB
    aws_metric_name: HTTPCode_Backend_5XX
    aws_dimensions: [LoadBalancerName]
    aws_statistics: [Sum]

Verify these values:

• <aws-region>: Your AWS region where ELBs are deployed (e.g., us-east-1, eu-west-1, ap-south-1).

See example configurations for more service templates.

Step 2: Download and Run the Exporter

curl -LO https://repo1.maven.org/maven2/io/prometheus/cloudwatch/cloudwatch_exporter/0.16.0/cloudwatch_exporter-0.16.0-jar-with-dependencies.jar

java -jar cloudwatch_exporter-0.16.0-jar-with-dependencies.jar 9106 elb-metrics.yaml

docker run -d \
  --name cloudwatch-exporter \
  -p 9106:9106 \
  -v $(pwd)/elb-metrics.yaml:/config/config.yml \
  -e AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY_ID \
  -e AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY \
  prom/cloudwatch-exporter

Verify the exporter is running by checking the metrics endpoint:

curl http://localhost:9106/metrics | grep aws_elb

You should see metrics like aws_elb_request_count_sum, aws_elb_latency_average, etc.

Step 3: Configure OpenTelemetry Collector

Add the following prometheus receiver to your existing otel-collector-config.yaml to scrape the CloudWatch Exporter:

receivers:
  prometheus:
    config:
      scrape_configs:
        - job_name: 'elb-cloudwatch'
          scrape_interval: 60s
          static_configs:
            - targets: ['<exporter-host>:9106']

Verify these values:

• <exporter-host>: The hostname or IP where the CloudWatch Exporter is running. Use localhost if running on the same machine.

Enable the prometheus receiver in your metrics pipeline by updating the service section:

service:
  pipelines:
    metrics:
      receivers: [otlp, prometheus]
      processors: [batch]
      exporters: [otlp]

Restart your OpenTelemetry Collector to apply the changes.

Validate

To confirm that ELB metrics are flowing to SigNoz:

1. Navigate to Dashboards → New Dashboard → New Panel in SigNoz.
2. In the query builder, search for metrics starting with aws_elb_ (e.g., aws_elb_request_count_sum).
3. Verify that metrics appear with load_balancer_name labels matching your ELBs.

If you see metrics with these labels, your setup is working correctly.