Simple Load Testing with curl

Recently I was testing a client's rate limiting configuration, so I needed to send a bunch of requests to a URL within a short period of time to see if it was working properly. I wrote up a quick bash script to loop from 1 to 20 with curl requests and sleeping for a short period between each request.

I thought it would be pretty easy to make the url, number of requests and delay command line arguments. It was, and I now have a handy general purpose tool for making repeated requests with curl. You can run the tool like this:

load.sh http://example.com/ 5 1

And it will make 5 requests to example.com with a 1 second delay between each http request. Here's the output

Hitting http://example.com with 5 requests with 0.5 delay
Response 1 200 0.207889s
Response 2 200 0.184135s
Response 3 200 0.188796s
Response 4 200 0.100910s
Response 5 200 0.097209s
5/5 Sent

The script will output the http response status code and the time taken for each request.

Here's the bash script, I call load.sh:

#!/bin/bash

if [ $# -lt 1 ]; then
    echo "Usage: $0 url [requests] [delay]"
    exit 1
fi

test_url="$1"
requests="${2:-1}"
delay="${3:-0}"

if ! [[ "$requests" =~ ^[0-9]+$ ]] || [ "$requests" -lt 1 ] || [ "$requests" -gt 1000 ]; then
    echo "Error: requests must be an integer between 1 and 1000"
    exit 1
fi

echo "Hitting $test_url with $requests requests with $delay delay"

for ((num=1;num<=$requests;num++ ))
do
	curl $test_url -o /dev/null -w "Response $num %{http_code} %{time_total}s\n" -s &
	sleep $delay
done

echo "$requests/$requests Sent"

wait

You'll want to save the above as load.sh then chmod a+x load.sh.

The first part of the script is really just putting the arguments into variables and doing a quick check on the number of requests to make sure you don't create an infinite loop by accident. Then we have a bash for loop that goes from 1 to the number of requests specified. For each iteration of the loop curl is invoked, and I use the ampersand & at the end to put the curl process in the background. This is the entire key to the script working, it allows you to send concurrent requests, and acheive a somewhat steady request rate. Finally at the end of the script I call wait which waits for my background commands to finish before exiting the program.

I wrote and tested this script on a Mac, but it should also work just fine on Linux or on Windows with WSL.