server : Add k6 Load Testing Script (#3175)

* add load testing script and update README for k6 integration

diff --git a/examples/server/README.md b/examples/server/README.md
index 4f2e40364cdc0cc78c221b891b37d12ed37346d1..c8e2f714bceef449dc5758d33a616b36b6e7a411 100644 (file)
--- a/examples/server/README.md
+++ b/examples/server/README.md
@@ -67,3 +67,35 @@ curl 127.0.0.1:8080/load \
 -H "Content-Type: multipart/form-data" \
 -F model="<path-to-model-file>"
 ```
+
+## Load testing with k6
+
+> **Note:** Install [k6](https://k6.io/docs/get-started/installation/) before running the benchmark script.
+
+You can benchmark the Whisper server using the provided bench.js script with [k6](https://k6.io/). This script sends concurrent multipart requests to the /inference endpoint and is fully configurable via environment variables.
+
+**Example usage:**
+
+```
+k6 run bench.js \
+  --env FILE_PATH=/absolute/path/to/samples/jfk.wav \
+  --env BASE_URL=http://127.0.0.1:8080 \
+  --env ENDPOINT=/inference \
+  --env CONCURRENCY=4 \
+  --env TEMPERATURE=0.0 \
+  --env TEMPERATURE_INC=0.2 \
+  --env RESPONSE_FORMAT=json
+```
+
+**Environment variables:**
+- `FILE_PATH`: Path to the audio file to send (must be absolute or relative to the k6 working directory)
+- `BASE_URL`: Server base URL (default: `http://127.0.0.1:8080`)
+- `ENDPOINT`: API endpoint (default: `/inference`)
+- `CONCURRENCY`: Number of concurrent requests (default: 4)
+- `TEMPERATURE`: Decoding temperature (default: 0.0)
+- `TEMPERATURE_INC`: Temperature increment (default: 0.2)
+- `RESPONSE_FORMAT`: Response format (default: `json`)
+
+**Note:**
+- The server must be running and accessible at the specified `BASE_URL` and `ENDPOINT`.
+- The script is located in the same directory as this README: `bench.js`.

diff --git a/examples/server/bench.js b/examples/server/bench.js
new file mode 100644 (file)
index 0000000..1275443
--- /dev/null
+++ b/examples/server/bench.js
@@ -0,0 +1,29 @@
+import http from 'k6/http'
+import { check } from 'k6'
+
+export let options = {
+  vus: parseInt(__ENV.CONCURRENCY) || 4,
+  iterations: parseInt(__ENV.CONCURRENCY) || 4,
+}
+
+const filePath        = __ENV.FILE_PATH
+const baseURL         = __ENV.BASE_URL        || 'http://127.0.0.1:8080'
+const endpoint        = __ENV.ENDPOINT        || '/inference'
+const temperature     = __ENV.TEMPERATURE     || '0.0'
+const temperatureInc  = __ENV.TEMPERATURE_INC || '0.2'
+const responseFormat  = __ENV.RESPONSE_FORMAT || 'json'
+
+// Read the file ONCE at init time
+const fileBin = open(filePath, 'b')
+
+export default function () {
+  const payload = {
+    file:           http.file(fileBin, filePath),
+    temperature:    temperature,
+    temperature_inc: temperatureInc,
+    response_format: responseFormat,
+  }
+
+  const res = http.post(`${baseURL}${endpoint}`, payload)
+  check(res, { 'status is 200': r => r.status === 200 })
+} 
\ No newline at end of file