Nodes always publish container port 25565, so an owner-set SERVER_PORT made Java Edition unreachable. The image already defaults to 25565; the status ping uses pingPort 25565 instead of reading the setting.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Nodes always publish container port 7777, so an owner-set Port made the server unreachable. The startup command now passes Port=7777 directly.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
First boot with the install fix ran ARK for ~10 minutes and found, in
ShooterGame.log: "Server: ... has successfully started!", "EOS ownership
validation initialized: Success", "Full Startup: 36.26 seconds" and,
about two minutes later, "Server has completed startup and is now
advertising for join. (9.96GB Mem)". That last line is when players,
console players through EOS included, can join, so it is the
readyPattern; the real line is test.yaml's readyExample. Boot tests now
wait for it instead of only checking ARK stays up for 10 minutes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The image's start_server runs SteamCMD once and carries on even when it
fails; SteamCMD's intermittent "Missing configuration" error then left
no server installed, so ARK exited and only the node's restart policy
saved it (a boot test would fail on the restart). First boot showed it.
The startup command now downloads SteamCMD if missing, runs the usual
"+login anonymous +quit" warm-up and installs app 2430930, retrying up
to three times until ArkAscendedServer.exe exists. start_server's own
install step then only validates.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ARK's image is built around user gameserver (uid 25000) and locks
/home/gameserver to it, so under the node's default 1000:1000 the
startup script got "Permission denied" and the server restart-looped,
on real nodes as well as in boot tests. Set containerUser 25000:25000
(needs node-agent and panel support for containerUser).
The startup command read ${RCON_PORT}, but the node passes the RCONPort
setting as RCONPORT, so custom RCON ports were ignored for 27020.
Boot-tested in the Linux agent image: up 30 minutes with no restart.
The test now waits 10 minutes, covering the first ~13 GB download.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With MEMORY=4G and requiredMemoryMb 4096, Java's heap plus overhead
exceeded the container limit and the server was killed on start; the
boot test failed every time. With 3G it is ready in about 18s at 4 GB.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Only games whose folder changed are boot-tested (against the previous
main commit, or against main on branches; manual runs pick games). Each
one is deployed by the node-agent's real deploy code on a dedicated
game-test runner, in a network that can reach the internet but not the
LAN. Publishing to the panel runs only if they all booted.
- boottest-payload builds the DEPLOY_SERVER message the panel would send
- readyPattern/failPatterns in game.yaml; readyExample in test.yaml is
checked against the pattern during validation
- secrets in test.yaml come from CI; a game missing one is skipped
- Minecraft gets a ready pattern; Dragonwilds needs DRAGONWILDS_OWNER_ID
- RUNNER.md + scripts/runner: runner, isolated network, systemd unit
Also fixes newline values ("\n" separators/footers) that yaml.v3 wrote
as "" on import, which would have broken ARK and Dragonwilds config
files once synced, and adds a round-trip test for the importer.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Public game templates for GBR Hosting, one games/<id>/game.yaml per game,
imported from the panel. A Go tool validates them against JSON Schemas and
checks for credentials, private addresses and consents that default to
accepted; CI runs it on every push.
Admin/RCON passwords are no longer shared defaults: they are blank with
x-generate so the panel creates one per server, and Palworld/ARK no longer
publish their RCON/REST ports.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
greg
