Migrating from docker-compose

If you’re currently running docker compose up -d on a single host, you can drop RunWisp in front without rewriting a line of YAML. Every compose service becomes an observable RunWisp service: logs streamed and indexed, restart policies enforced, notifications on failure, trigger/stop from the Web UI and REST API. The containers themselves still run under Docker, exactly as before — RunWisp is the supervisor and observability layer on top.

60-second migration

1. Drop the runwisp binary next to your docker-compose.yml.

2. Run runwisp daemon. On first run, RunWisp detects the compose file and scaffolds a starter runwisp.toml that imports every service:

   [compose.myapp]
   file = "./docker-compose.yml"

3. Open the Web UI on :8080 (default). Every compose service is a RunWisp service tagged with a small “compose” badge.

That’s the whole migration. From here, layer on the RunWisp knobs you want without touching the compose file.

Common follow-ups

Restart everything on failure

Compose-imported services restart on_failure by default. To make critical ones restart unconditionally:

[compose.myapp.api]
restart = "always"

Notify on failure

[[notifier]]
id    = "slack-prod"
kind  = "slack"
url   = "https://hooks.slack.com/services/..."

[compose.myapp.api]
notify_on_failure = ["slack-prod"]

Scale a worker horizontally

[compose.myapp.worker]
instances = 4

Four containers named myapp_worker_0 … myapp_worker_3 come up, each receiving a distinct RUNWISP_INSTANCE_INDEX env var (0..3) so they can self-shard.

Run only a subset of services through RunWisp

[compose.myapp]
include = ["api", "worker"]   # leaves `db` etc. unmanaged

…or:

[compose.myapp]
exclude = ["db"]

Trigger compose images as cron jobs

A nightly backup using a compose-declared image, without bringing the container up as a long-running service:

[tasks.nightly-backup]
cron            = "0 3 * * *"
compose_file    = "./docker-compose.yml"
compose_service = "backup"

What stays under your control

• The compose file itself — networks, volumes, depends_on, image build contexts, healthchecks. RunWisp reads it; it never rewrites it.
• The Docker daemon — RunWisp shells out to docker compose; install Docker the way you already do.
• The application — RunWisp doesn’t proxy traffic, doesn’t intercept stdin, doesn’t change container networking. docker compose ps, docker logs, and docker exec all keep working as before.

What RunWisp adds

• Per-run history — every container start is a RunWisp run with a ULID, start/end timestamps, exit code, and captured stdout/stderr.
• Restart policy — RunWisp owns supervision now; the compose-file restart: directive is ignored on imported services.
• Notifications — wire compose services into the same notification rules used for [tasks.*] and [services.*].
• Manual trigger/stop — start and stop containers from the Web UI or REST API. Stops are graceful; the compose-declared stop_grace_period is honoured.

See [compose.*] for the full reference.