How to execute multiple queries in one call in Prometheus

In Prometheus, executing multiple queries in a single operation can be achieved through batch queries using the API. The Prometheus HTTP API provides a mechanism that allows users to execute multiple queries simultaneously and retrieve their results programmatically. Below, I will detail how to perform this operation using the API.

Step 1: Constructing the API Request

First, you need to construct a request to the Prometheus HTTP API. The primary parameters required for each query are:

• query: Query expression.
• time: Query timestamp.
• timeout (optional): Query timeout.

For example, if we want to query both the system's CPU usage and memory usage concurrently, we can construct the following requests:

bash
curl -G -s 'http://<prometheus_host>:9090/api/v1/query' --data-urlencode 'query=rate(node_cpu_seconds_total[5m])' --data-urlencode 'time=1609459200'
curl -G -s 'http://<prometheus_host>:9090/api/v1/query' --data-urlencode 'query=node_memory_MemFree_bytes' --data-urlencode 'time=1609459200'

Step 2: Sending Multiple Requests in Parallel

You can use a Bash script or any programming language that supports HTTP requests to send these queries in parallel. The following is an example script using Bash and curl, which executes multiple Prometheus queries concurrently:

bash
#!/bin/bash

# Define queries and time
queries=("rate(node_cpu_seconds_total[5m])" "node_memory_MemFree_bytes")
time="1609459200"

# Execute all queries in parallel
for query in "${queries[@]}"
do
  curl -G -s "http://<prometheus_host>:9090/api/v1/query" --data-urlencode "query=$query" --data-urlencode "time=$time" &

done

wait

Step 3: Parsing and Using Results

Results will be returned in JSON format. Each request's response contains the query data. You can parse these JSON responses to further process or display the results.

Notes

• Ensure that the Prometheus server address and port are correctly configured.
• Set the query time according to actual requirements.
• If high performance is needed or the query volume is large, consider distributed queries or enhancing the performance of the Prometheus server.

By following these steps, you can effectively execute multiple Prometheus queries in a single operation, significantly improving the efficiency of data retrieval and monitoring.