Running the daemon

overcast serve runs the emulator in the foreground. overcast start runs the same daemon detached and records it under a name, so stop, restart and logs can find it again.

overcast serve                      # foreground; Ctrl+C stops it
overcast start --name ci --no-wait  # detached, recorded as "ci"

Part of the CLI reference.

overcast serve

Emulator configuration is by environment variable — see the environment variable reference. serve itself takes three flags, for the web console and the optional mDNS bridge.

overcast serve
OVERCAST_STATE=memory overcast serve
overcast serve --bridge

Background instances

overcast start spawns overcast serve as a detached process — or a container, with --docker — and records it at ~/.overcast/instances/<name>/. stop, restart and logs look the instance up there by name, and overcast status lists every record. A foreground overcast serve is in no registry and none of them can see it.

overcast start

overcast start                              # default instance on 4566/4567
overcast start --name ci --port 4570 --ui-port 0 --no-wait
overcast start --docker --channel latest --mount-docker-socket

Given neither --image nor --channel, a --docker instance runs ghcr.io/overcast-sh/overcast at this CLI’s own version tag rather than a floating one, so the container always matches the binary that launched it. An unreleased (dev) build has no matching tag and falls back to :alpha, saying so as it starts.

Starting an instance whose name is already running fails with a pointer to overcast stop; a dead record left by a crashed process is replaced silently. On success start prints the endpoint, the web console URL, where the log lives, and the overcast env line that points AWS tools at it.

overcast stop [name]

Stops the named instance (default if omitted) and removes its registry record. A native process is asked to exit and killed if it has not gone after 10 seconds; a container gets docker stop then docker rm. Stopping an already-stopped instance is not an error — it clears the stale record.

overcast stop
overcast stop ci

overcast restart [name]

Stops the instance if it is running, then starts it again from its saved configuration — the flags the original overcast start was given, replayed without your having to remember them. Takes the same --no-wait and --timeout flags as start.

overcast restart
overcast restart ci --no-wait

overcast logs [name]

Shows a background instance’s output: the daemon.log file for a native instance, docker logs for a container. Same flags either way.

overcast logs
overcast logs ci --follow --tail 500

Related

• Checking a running daemon — status, wait, services, config
• Environment variable reference — every environment variable serve reads
• Storage and persistence — what --state and --data-dir choose between