How to Conduct Accurate API Load Tests with Apipost’s One‑Click Stress Tool

Overview

Apipost provides an integrated platform for API documentation, debugging, mock, and testing. It supports common HTTP methods such as POST, GET, and PUT, and can generate and export API documentation.

Background

After an API has been debugged, verifying its performance under high‑concurrency load is essential. This typically requires coordination between development and testing teams. Apipost’s one‑click stress test offers a streamlined way to perform such verification.

Factors Influencing Load Test Results

External conditions can distort concurrency metrics. Key influences include:

Local file‑handle (descriptor) limits – the default OS limit is often 1024.

DNS resolution latency.

Network quality.

Server‑side connection limits.

For instance, attempting 1w (10,000) concurrent requests usually exceeds the file‑handle limit, causing connection failures.

Recommended Concurrency Strategy

Test with gradually increasing concurrency while monitoring the failure rate. A typical progression is:

10 concurrent requests

100 concurrent requests

500 concurrent requests

1,000 concurrent requests

If the failure rate stays below 1 %, increase the concurrency further. A healthy system shows a sustained rise in requests per second as concurrency grows.

Using the One‑Click Stress Test

Select the target endpoint in the API list.

Click the “One‑Click Stress Test” button.

Configure the desired concurrency level.

Choose a stress‑test mode (duration‑based or round‑based) and enter the corresponding value.

Start the test.

Note: The stress‑test service must be enabled before use.

Result Generation

When the test finishes, Apipost generates a log file that records each round’s metrics, including timestamp, sequence number, worker ID, HTTP status code, response time, and message. Example entries:

17:18:30 seq:6 work:0 code:200 cost:66.0 msg:ok
17:18:30 seq:9 work:3 code:200 cost:68.0 msg:ok
17:18:30 seq:2 work:5 code:200 cost:72.0 msg:ok
…
17:18:40 seq:818 work:3 code:200 cost:110.0 msg:ok
17:18:40 seq:819 work:4 code:200 cost:93.0 msg:ok
17:18:40 seq:816 work:5 code:200 cost:143.0 msg:ok

Result Analysis

During execution the interface shows “requests per second”. After completion a report is produced, summarizing total requests, average latency, success rate, and any errors. The calculation follows the standard formula:

Requests per second = total successful requests ÷ test duration

Server‑Side Considerations

Ensure the server can accept the configured number of concurrent connections. Adjust OS limits (e.g., increase ulimit -n on Linux) and verify that the application’s thread pool and socket settings are sufficient.

Further Reference

Detailed documentation is available at: https://wiki.apipost.cn/docs/%E6%9C%80%E4%BD%B3%E5%AE%9E%E8%B7%B5/%E6%8E%A5%E5%8F%A3%E5%8E%8B%E6%B5%8B%E5%AE%9E%E8%B7%B5