Demos & training

01

The quickstart 0:24

A complete load test in 20 lines of YAML: validate it, run it, read the summary.

▸ the scenarios / flow / thresholds structure
▸ loadr validate before you burn a test run
▸ live progress, then the k6-style summary
▸ checks, latency percentiles and green thresholds

Docs: Your first test

02

Validation that respects you 0:16

Broken test in, precise diagnostics out — before a single request is sent.

▸ line & column on every finding
▸ did-you-mean fixes for typos
▸ executor parameter checks
▸ undefined-variable detection

Docs: loadr validate

03

Data, extraction & JS in one flow 0:18

The correlation workflow every real test needs.

▸ CSV rows feeding each login
▸ JSONPath-extracted token replayed via ${token}
▸ inline ${js: crypto.uuidv4()}
▸ custom metrics with exact-count thresholds

Docs: Extraction · Data · JavaScript

04

Thresholds gate your CI 0:20

What failure looks like — and how it reaches your pipeline and your team.

▸ an SLO the endpoint can't meet
▸ red ✗ thresholds, exit=99
▸ --summary-export for machines
▸ loadr report HTML for humans

Docs: Thresholds · Exit codes

05

Escape JMeter. Import k6. 0:21

Bring your existing test estate with you.

▸ a JMeter .jmx plan → clean YAML
▸ warnings name exactly what to review
▸ a k6 script with scenarios → same
▸ converted output always validates

Docs: From JMeter · From k6

06

A distributed fleet, live 0:34

Controller, agents, exact partitioning and merged percentiles — in 30 seconds.

▸ loadr controller + two joining agents
▸ fleet health from the API
▸ 400 iterations split exactly in two
▸ HDR-merged, fleet-wide percentiles

Docs: Distributed testing · Metric aggregation

07

Flow control & weighted tasks 0:23

Loops, conditionals and weighted branches — Gatling's DSL and Locust's task model in YAML.

▸ while loop: browse a random number of pages
▸ weighted random: 70% browse / 25% cart / 5% checkout
▸ repeat and if/else for retries
▸ realistic user journeys, no scripting required

Docs: Flow control

08

Feeders & throttling 0:21

Random data feeders and a hard request-rate ceiling — Gatling's feeders and reachRps.

▸ a random feeder picking SKUs
▸ 20 VUs, but throttle caps the scenario at 30 req/s
▸ the observed rate holds at the ceiling
▸ stay under a known rate limit, on any executor

Docs: Feeders & throttling

09

The live web UI 0:42

A real browser session during a 30-VU run.

▸ live RPS, VU and percentile charts (SSE)
▸ threshold PASS pills updating in real time
▸ run page with Stop / Kill / Pause controls
▸ test editor, agents view, dark mode native

Docs: The management UI

10

The agent fleet, in the UI 0:37

A controller with three live agents, managed from the browser — the distributed run from #6, seen through the web UI.

▸ the fleet view: three healthy agents, live
▸ 30 VUs partitioned exactly 10 / 10 / 10
▸ per-agent cores, labels and heartbeat
▸ the overview shows centrally-merged percentiles

Docs: Controller & agents

11

The time-series HTML report 0:28

A ramp-to-spike run, charted over time — one self-contained HTML file, no servers, no CDN.

▸ throughput, latency p50/p95/p99, VUs & error rate over time
▸ a shared hover crosshair reads off every chart at once
▸ the spike in load, mirrored in the latency curve
▸ exact aggregate tables below — one file you can email

Docs: HTML reports

12

JMESPath & fused check-chains 0:09

Extract, coerce a type, transform and validate a value in one declarative step — Gatling's check chains, in YAML.

▸ a jmespath filter + projection over the JSON body
▸ as: int type coercion and a transform pipeline
▸ an inline check that validates before saving
▸ chains and classic extractors mixed in one request

Docs: Extraction & check-chains

13

Failure breakdown, by cause 0:30

A run that fails on purpose — 4xx/5xx, a failing check and a JS exception — grouped by cause in the UI, then exported to CSV from the browser.

▸ failures grouped by HTTP status, transport error, check and exception
▸ per-cause counts and share-of-total bars, live over SSE
▸ normalised exception messages so the same bug groups together
▸ one-click CSV / HTML report download, entirely in the browser

Docs: The management UI

14

Load-test SQL databases 0:20

Drive PostgreSQL and MySQL directly — the query is the request, against real databases.

▸ postgres:// and mysql:// targets, one query per request
▸ parameterised SELECT/INSERT, bound safely by the driver
▸ per-VU connection pooling via sqlx
▸ sql_req_duration, sql_rows metrics + thresholds

Docs: SQL (PostgreSQL & MySQL)

15

Install the MongoDB plugin 0:24

Install a runtime protocol plugin, then drive MongoDB directly — insert, find and aggregate against a real database.

▸ loadr plugin install mongo — sha256-verified, ABI-checked
▸ a mongodb:// request routes straight to the plugin
▸ insert / find / aggregate operations, one per request
▸ mongo_reqs, mongo_req_duration, mongo_docs metrics

Docs: MongoDB plugin · Plugins

16

Install the PostgreSQL plugin 0:24

Pull the PostgreSQL driver at runtime, then drive parameterised queries against a real database — the query is the request.

▸ loadr plugin install postgres — driver loads at runtime
▸ postgres:// targets, parameterised SELECT/INSERT
▸ $1, $2 placeholders bound safely, pooled per VU via sqlx
▸ postgres_reqs, postgres_req_duration, postgres_rows

Docs: PostgreSQL plugin · Plugins

17

Install the MySQL plugin 0:24

Install the MySQL driver on demand, then run a steady query workload against a real database over the wire protocol.

▸ loadr plugin install mysql — never baked into the binary
▸ mysql:// targets over the MySQL wire protocol
▸ ? placeholders bound safely, pooled per VU via sqlx
▸ mysql_reqs, mysql_req_duration, mysql_rows

Docs: MySQL plugin · Plugins

18

Install the Redis plugin 0:24

Install a runtime protocol plugin, then drive Redis directly over RESP — SET, GET and INCR against a real cache.

▸ loadr plugin install redis — sha256-verified, ABI-checked
▸ a redis:// request routes straight to the plugin
▸ one RESP command per request — SET / GET / INCR
▸ redis_reqs, redis_req_duration metrics

Docs: Redis plugin · Plugins

19

Install the Kafka plugin 0:24

Pull the Kafka driver at runtime, then produce and fetch messages against a real broker — a pure-Rust client, no librdkafka.

▸ loadr plugin install kafka — pure-Rust rskafka client
▸ kafka://broker:9092/topic — broker authority, topic path
▸ produce and fetch operations, one per request
▸ kafka_reqs, kafka_req_duration, kafka_msgs

Docs: Kafka plugin · Plugins

20

Install the RabbitMQ plugin 0:24

Install the AMQP driver on demand, then publish and drain a queue against a real RabbitMQ broker — the pure-Rust lapin client.

▸ loadr plugin install rabbitmq — pure-Rust lapin client
▸ amqp:// / amqps:// targets, AMQP 0.9.1
▸ publish and get operations, queues declared inline
▸ rabbitmq_reqs, rabbitmq_req_duration, rabbitmq_msgs

Docs: RabbitMQ plugin · Plugins

21

Install the Elasticsearch plugin 0:24

Pull the Elasticsearch driver at runtime, then index, bulk-load and search against a real cluster over HTTP/JSON.

▸ loadr plugin install elasticsearch — over hyper + hyper-rustls
▸ elasticsearch:// / es:// targets, plain HTTP/JSON
▸ index / get / search / bulk operations, one per request
▸ elasticsearch_reqs, elasticsearch_req_duration, elasticsearch_docs

Docs: Elasticsearch plugin · Plugins

22

Write a plugin in Go 0:34

loadr plugins aren't Rust-only. Build a protocol plugin in Go with go build -buildmode=c-shared, install it, and drive it with the real binary — loaded over loadr's frozen C ABI.

▸ a Go shared library, not Rust — built with go build -buildmode=c-shared
▸ exports four extern "C" symbols over loadr's frozen C ABI
▸ serves the goecho:// scheme; the host auto-detects the C ABI at load
▸ goecho_reqs — status 200 + body assertions pass, 0% failed

Docs: Plugins in other languages (C ABI) · source: go-echo

Watch loadr work.