Local (Docker)

The local deployment uses Docker Compose to run Interlock with Redis and Postgres. This is ideal for development, testing, and demo purposes.

Services

The demo/local/docker-compose.yml defines:

Service	Image	Purpose
`redis`	`redis:7-alpine`	Primary state store
`postgres`	`postgres:16-alpine`	Archival storage
`seed`	Custom (Go)	Registers pipelines, serves webhook endpoint
`interlock`	Custom (Go)	Runs the watcher loop

Seed Container

The seed container registers pipelines and archetypes at startup, then exposes a webhook server on port 8888 for HTTP triggers to call.

Interlock Container

The interlock container runs the watcher loop. It connects to Redis for state and Postgres for archival. Evaluator scripts are mounted as volumes and run as subprocess evaluators.

Running Locally

cd demo/local
docker compose up -d

To rebuild after code changes:

docker compose up -d --build

Evaluators

The local demo uses shell-script evaluators that read sensor values from Redis:

check-freshness

Reads sensor:freshness from Redis. Returns PASS if the value is below the configured threshold.

#!/bin/bash
VALUE=$(redis-cli -h redis GET sensor:freshness)
if [ "$VALUE" -lt "$THRESHOLD" ]; then
  echo '{"status":"PASS","value":{"minutes":'$VALUE'},"reason":"data is fresh"}'
else
  echo '{"status":"FAIL","value":{"minutes":'$VALUE'},"reason":"data is stale"}'
fi

check-record-count

Reads sensor:record-count from Redis. Returns PASS if count meets the minimum.

check-upstream

Reads sensor:upstream from Redis. Returns PASS if the upstream pipeline completed.

Setting Sensor Values

Control evaluator outcomes by writing sensor values directly to Redis:

docker compose exec redis redis-cli SET sensor:freshness 15
docker compose exec redis redis-cli SET sensor:record-count 5000
docker compose exec redis redis-cli SET sensor:upstream completed

E2E Test Suite

The local E2E tests (demo/local/e2e-test.sh) run 6 scenarios that mirror the AWS E2E suite:

Scenarios

#	Scenario	Description
1	Progressive readiness	Traits start failing, sensors updated to passing, pipeline triggers
2	Quality drop re-run	Pipeline completes, quality drops, re-evaluation triggers re-run
3	Dedup	Same pipeline+date+schedule evaluates only once
4	Excluded day	Pipeline with exclusion on current day is skipped
5	Pipeline not found	Non-existent pipeline returns error gracefully
6	Watcher-driven	Full watcher loop integration (register → evaluate → trigger → complete)

Running E2E Tests

make local-e2e-test

This runs the full suite and saves results to demo/local/e2e-results/.

Cleanup and Teardown

Between scenarios, run logs are cleaned up to ensure isolation:

docker compose exec redis redis-cli DEL interlock:runlog:my-pipeline:2026-02-23:daily

Full teardown:

make local-e2e-test-teardown

This runs docker compose down -v to remove containers and volumes.

Archival Verification

The E2E suite verifies Postgres archival:

Archiver interval is set to 30 seconds for testing
After scenarios complete, wait 35 seconds for archival
Assert rows exist in runs and events Postgres tables

docker compose exec postgres psql -U interlock -d interlock \
  -c "SELECT COUNT(*) FROM runs;"

Configuration Files

The local demo mounts configuration from demo/local/:

demo/local/
├── docker-compose.yml
├── interlock.yaml          # Watcher configuration
├── pipelines/              # Pipeline YAML definitions
├── archetypes/             # Archetype YAML definitions
├── evaluators/             # Shell script evaluators
├── calendars/              # Calendar YAML files
├── e2e-test.sh             # E2E test runner
└── e2e-results/            # Test output (gitignored)

Terraform (AWS)