Distributed fleet

Scale & operations

Scale past one box: a controller splits the load across a fleet of agents and merges their HDR histograms into one true result — accurate percentiles, not an average of averages.

Browse all 52 demos 12 categories
Run it
$ loadr run examples/15-distributed.yaml
examples/15-distributed.yaml
# Distributed run: submit this to a controller and the load is partitioned
# across every connected agent (VUs split, arrival rates divided, percentiles
# merged centrally from HDR histograms — never averaged).
#
#   loadr controller --bind 0.0.0.0:7625 &
#   loadr agent --join controller-host:7625 --name agent-1 &
#   loadr run --controller controller-host:6464 examples/15-distributed.yaml
name: distributed
description: 600 req/s split across the agent fleet

defaults:
  http:
    base_url: https://api.example.com

scenarios:
  fleet_load:
    executor: constant-arrival-rate
    rate: 600                 # total across ALL agents
    duration: 15m
    pre_allocated_vus: 300
    max_vus: 900
    flow:
      - request:
          name: search
          url: "/search?q=stress"
          checks: [ { type: status, equals: 200 } ]

  # Scale this one live from the web UI's run page or the controller API.
  manual_dial:
    executor: externally-controlled
    max_vus: 500
    duration: 15m
    flow:
      - request: { name: home, url: / }

thresholds:
  http_req_duration: [ "p(95)<400", "p(99)<900" ]
  http_req_failed: [ "rate<0.01" ]

View raw: examples/15-distributed.yaml

What it shows

  • Controller + agent fleet
  • HDR histogram merge
  • One plan, many machines