Demo compose¶

This compose runs Tapyr against the simulators of the built-in providers, without any vendor account. Its services:

docker compose up -d
docker compose logs -f tapyr     # JSON logs
docker compose down -v           # stop and forget the data

The first samples arrive within a minute (the demo spreads the first runs over 1% of their interval instead of the default 10%). The ports listen on 127.0.0.1 only.

The simulators listen at 8081 plus the portOffset of their package (sim/sim.yaml): aruba-central-classic 8081, example-sim 8082, meraki-dashboard 8083, teltonika-rms 8084, teltonika-rutos 8085. The credentials in compose.yaml are the ones the simulators accept; none of them is a real secret.

Try the fault scenarios¶

Every provider package has a quota-tight scenario (a quota smaller than the plan expects, or another client consuming it) and a bad-day scenario (errors, latency, an outage). Start the simulators with one of them and watch Tapyr adapt in the Budgets and Plan pages of the UI and in the Quotas row of the overview dashboard:

SIM_SCENARIO=quota-tight docker compose up -d sim
SIM_SCENARIO=bad-day docker compose up -d sim
docker compose up -d sim           # back to the default behavior

A scenario is a timeline that starts with the simulators. quota-tight acts at once. bad-day waits an hour before its first fault, then goes through added latency, errors on the main routes and a short outage, at times of their own in every package (sim/scenarios/bad-day.yaml): the outage of RutOS after 1 hour, the errors after 2 hours, the outages of Meraki and RMS after 4 hours. To see it sooner, run the simulators faster than real time:

SIM_SCENARIO=bad-day SIM_TIME_SCALE=10 docker compose up -d sim

The scenarios then play ten times faster: the outage of RutOS after 6 minutes, the errors after 12, the other outages after 24. The faults get shorter as well, and the quota windows of the simulators run faster too: well above 10, most faults end before the jobs of the demo run again, and Tapyr hardly sees them. Retries absorb most of the errors: look for them on the Retries per minute, by reason panel of the Tapyr / Overview dashboard rather than in the logs.

With your own build¶

The compose uses the image of the release. To try a local build:

make docker                                   # from the root of the repository: tapyr:dev
TAPYR_IMAGE=tapyr:dev docker compose up -d