Prometheus is a versatile tool for monitoring and alerting that stores and manages time-series data. When querying Prometheus using its Query Language (PromQL), you may encounter situations where your queries contain curly braces {}. One common misconception is that you can use GET requests for these queries. However, GET requests may not handle curly braces as expected, leading to confusion. In this blog post, we’ll explain why this happens and how to work around it using POST requests in Python.

Understanding the Issue Curly braces are an integral part of PromQL queries. They are used to filter time-series data based on labels, such as selecting metrics with specific labels or values. For example, a PromQL query to retrieve metrics for HTTP requests with a status code of 200 might look like this:

promQL Copy code http_requests_total{status_code=“200”} When you send this query as a GET request, you might expect to receive the relevant data. However, using GET requests for these types of queries often results in unexpected behavior because the curly braces are URL-encoded.

Why GET Requests Fail GET requests are designed to send data in the URL’s query parameters. When you include curly braces in the URL, they are encoded as %7B for { and %7D for }. For the example query above, the URL-encoded GET request would look like:

perl Copy code http://your-prometheus-instance:9090/api/v1/query_range?query=http_requests_total%7Bstatus_code%3D%22200%22%7D&start=timestamp_start&end=timestamp_end&step=5m Prometheus, however, does not recognize this URL-encoded form as a valid PromQL query, and the query may fail or return unexpected results.

The Solution: Using POST Requests To overcome this issue, we can use POST requests with a JSON payload instead of GET requests. When you send the query in the request body as a JSON object, the curly braces remain intact and are not URL-encoded. Here’s an example of how to do this in Python using the requests library:

python Copy code import requests

Define Prometheus API endpoint

prometheus_url = 'http://your-prometheus-instance:9090'

Define your PromQL query with curly braces

promql_query = ‘http_requests_total{status_code=“200”}’

Define the time range and step interval

Build the API request

api_url = f'{prometheus_url}/api/v1/query_range'
payload = {
    'query': promql_query,
    'start': 'timestamp_start',
    'end': 'timestamp_end',
    'step': '5m'
}

# Send the POST request
response = requests.post(api_url, json=payload)

# Handle the response
if response.status_code == 200:
    query_result = response.json()
    # Process the query result as needed
else:
    print(f'Failed to execute query: {promql_query}')
In this example, we send the PromQL query with curly braces in the request body as a JSON object. This allows Prometheus to correctly interpret and execute the query.

Conclusion Using GET requests to query Prometheus with PromQL queries that contain curly braces is a common pitfall. To avoid issues with URL encoding and ensure that your queries are properly interpreted, it’s best to use POST requests with a JSON payload. By understanding this behavior and following the recommended approach, you can effectively query Prometheus data without unexpected results.