CheatVegeta HTTP load testing
Light · Postcard
102 × 152 mm · PDF
Dark · Postcard
102 × 152 mm · PDF
Ink saver · Postcard
102 × 152 mm · PDF
Light · A4 landscape
297 × 210 mm · PDF
Dark · A4 landscape
297 × 210 mm · PDF
Ink saver · A4 landscape
297 × 210 mm · PDF
Light · A3 landscape
420 × 297 mm · PDF
Dark · A3 landscape
420 × 297 mm · PDF
Ink saver · A3 landscape
420 × 297 mm · PDF
Start here
echo "GET <url>" | vegeta attack | vegeta
report
Quick test
vegeta attack -rate=<n> -duration=<t>
Fixed load
vegeta attack -targets=<file>
Many requests
vegeta attack > <file>.bin
Save results
vegeta report <file>.bin
Latency summary
vegeta report -type=json <file>.bin
Machine output
vegeta plot <file>.bin > <file>.html
Latency chart
vegeta encode -to=csv <file>.bin
Export results
Concepts
- Targets stdin or file
- Results binary stream
- Output text, JSON, HTML, CSV
Attack
vegeta attackFire requests from stdin targets
-rate=<n>/<unit>
Requests per time unit (default 50/1s)
-duration=<t>
How long to attack (0 = forever)
-targets=<file>
Read targets from a file, not stdin
-output=<file>
Write binary results to a file
-rate=0 -max-workers=<n>
Go as fast as possible
A rate of 0 or infinity needs a worker cap or it spawns
without limit.
-header="<Name: value>"
Add a request header (repeatable)
-body=<file>
Body file for every request
-timeout=<t>
Per-request timeout (default 30s)
-workers=<n>
Initial workers (default 10, grows)
-max-workers=<n>
Cap the worker pool
-name=<name>
Label the attack in results
-insecureIgnore invalid TLS certificates
-http2=falseForce HTTP/1.1
-keepalive=false
Open a new connection per request
-redirects=<n>
Max redirects (-1 = don't follow)
-format=jsonTargets are JSON lines
-cert <file> -key <file>
TLS client certificate and key
-root-certs <files>
Trusted root CAs, comma separated
-connect-to <src>:<dst>
Redirect host:port to another address
-resolvers <ip:port,...>
Use specific DNS servers
-dns-ttl=<t>
DNS cache time (-1 off, 0 forever)
-connections=<n>
Max idle connections per host
-lazyRead targets lazily to save memory
Saves memory with large target files but adds per-request
latency.
-chunkedSend body chunked
-h2cHTTP/2 without TLS
-unix-socket=<path>
Connect over a Unix socket
-laddr=<ip>
Local source IP address
-max-body=<size>
Cap captured response body bytes
-session-tickets
Resume TLS sessions
-prometheus-addr=<ip:port>
Expose live Prometheus metrics
The endpoint exists only while the attack runs.
Targets
# one request per block, blank line betweenGET http://localhost:8080/items?x=1X-Account-ID: 99POST http://localhost:8080/thingsContent-Type: application/json@./newthing.json{"method":"GET","url":"http://localhost/","header":{"X-Id":["1"]},"body":"<base64>"}Report
vegeta report <file>.bin
Text summary of an attack
cat <file>.bin | vegeta report
Read results from stdin
-type=jsonMetrics as JSON
-type=hist[0,100ms,200ms]
Latency histogram with buckets
-type=hdrplotHDR histogram plot data
-every=<t>
Print a report at each interval
vegeta report *.bin
Merge several runs by timestamp
| Field | Meaning |
|---|---|
| Requests | Total, rate sent, throughput |
| Latencies | Min, mean, p50/p90/p95/p99, max |
| Success | Share of 200-399 responses |
| Status Codes | Count per code; 0 = send failed |
| Error Set | Distinct errors seen |
Plot and encode
vegeta plot <file>.bin > <file>.html
Interactive latency-over-time chart
vegeta plot -title=<text>
Set the chart title
vegeta plot -threshold=<n>
Downsample above n points (4000)
vegeta plot a.bin b.bin
Compare runs in one chart
vegeta encode -to=csv <file>.bin
Convert results to CSV
The CSV has no header row.
vegeta encode -to=json <file>.bin
Convert results to JSON lines
vegeta encode -to=gob
Convert back to binary
Global
vegeta -versionPrint version
-cpus=<n>CPUs to use
-profile=cpu|heap
Profile vegeta itself
Recipes
echo "GET http://localhost:8080/" | \ vegeta attack -rate=100 -duration=10s | \ tee results.bin | vegeta reportecho "GET http://localhost/" > t.txtvegeta attack -targets=t.txt -name=50qps -rate=50 -duration=10s > r50.binvegeta attack -targets=t.txt -name=100qps -rate=100 -duration=10s > r100.binvegeta plot r50.bin r100.bin > plot.htmljq -ncM '{method:"POST", url:"http://localhost/", body:"{}"|@base64, header:{"Content-Type":["application/json"]}}' | \ vegeta attack -format=json -rate=100 -duration=5s | vegeta report# divide total rate by machines, run on each:vegeta attack -targets=t.txt -rate=250 -duration=60s > host1.bin# collect the files, then merge:vegeta report host*.bin
