Conversation
… links to didstopia npm server which wasn't responding
RustDedicated now requires GLIBC >= 2.34 (2.32/2.33/2.35 for UnityPlayer.so), but the ubuntu-20.04 base only ships glibc 2.31. The server binary fails to load immediately after install, causing containers to exit and restart-loop forever. Mirrors upstream Didstopia/rust-server, which moved to the nodejs-22-steamcmd-ubuntu-24.04 base and dropped the stale package-lock files (regenerated by npm install at build time). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…digest - Regenerate package-lock.json for all helper apps under node 22 instead of deleting them, and switch builds to npm ci for reproducible dependency resolution - Pin didstopia/base:nodejs-22-steamcmd-ubuntu-24.04 by manifest digest so rebuilds of the same commit get the same runtime (verified: image builds clean with these changes) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
fix: update base image to nodejs-22-steamcmd-ubuntu-24.04 (fixes GLIBC restart loop)
The healthcheck runs rcon against localhost:$RUST_RCON_PORT, but that env var only holds the image default (28016) — deployments that set the port via the bind-mounted /etc/rust/rust.env (sourced by start.sh) were always probed on the wrong port, so containers reported unhealthy even with the server fully up. Source rust.env in the healthcheck shell when present. Verified against a live server: probe now exits 0. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
fix: source rust.env in healthcheck so rcon probes the real RCON port
…is installed steamcmd exits 0 even when app_update fails. On 2026-09-03 the weekly and monthly servers restarted for their scheduled wipe at 18:00 UTC, the moment Facepunch's monthly depot was flipping; steamcmd printed "Error! App '258550' state is 0x486 after update job." and exited 0, the script carried on, installed an Oxide built for the new Rust (published 28 minutes earlier) over the old build, and RustDedicated hung at boot with MissingMethodException: Facepunch.ExceptionReporter.Initialize(string,string). A hang is not an exit, so `restart: unless-stopped` never fired; both servers stayed down for five days. install_or_update now accepts only steamcmd's own success line (also printed when already up to date). Anything else sleeps STEAMCMD_RETRY_DELAY (60s, so the retry loop does not hit Steam every ~20s — docker resets its backoff for a container that ran over 10s) and exits 1 before the Oxide step, so the restart policy retries until Steam serves the update. STEAMCMD_SH lets the new test stub the binary. test/install_or_update_test.sh extracts the function and drives it with a stub steamcmd printing the 0x486 line and the success line; the workflow runs it on PRs and gates the publish on it. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A39aALuGv7Pf3HCsbiqix8
Author
|
Opened against the wrong repository by mistake (meant for the compscidr fork). Sorry for the noise. |
There was a problem hiding this comment.
🟡 Changes recommended
The workflow/publish path and startup script have correctness/operational issues (missing GH Actions permissions, non-atomic date tagging, unguarded env sourcing noise) that should be fixed before merging.
Once you've addressed the issues Copilot identified, you can request another Copilot review.
Pull request overview
This PR hardens container startup so Rust is not started (and Oxide is not installed) when steamcmd fails to actually update Rust despite exiting 0, preventing a “new Oxide over old Rust” mismatch that can hang RustDedicated and defeat restart policies.
Changes:
- Add a guard in
install_or_updateto requiresteamcmd’sSuccess! App '258550' fully installedline; otherwise sleep and exit non-zero so restart policies retry safely. - Add a shell test to pin the failure mode and gate publishing via GitHub Actions.
- Update container build/runtime bits (base image/Node upgrade,
npm ci, lockfile updates) and document new behavior and env-file overrides.
File summaries
| File | Description |
|---|---|
start_rust.sh |
Enforces steamcmd success-line gating; adds env-file sourcing for overrides. |
test/install_or_update_test.sh |
Adds regression test for “steamcmd exited 0 but update failed” behavior. |
.github/workflows/docker-publish.yml |
Runs the new test on PRs and publishes from main. |
README.md |
Documents env-file overrides and steamcmd failure handling behavior. |
Dockerfile |
Updates base image/Node version and switches Node installs to npm ci; updates healthcheck env sourcing. |
docker-compose.yml |
Adds compose example including env/seed file mounts. |
rust.env |
Adds example rust env override file. |
seed.env |
Adds example seed override file. |
.gitignore |
Ignores IDE config and local rust-server data dir. |
shutdown_app/package-lock.json |
Lockfile v3 / dependency resolution updates. |
scheduler_app/package-lock.json |
Lockfile v3 format update. |
restart_app/package-lock.json |
Lockfile v3 / dependency resolution updates. |
rcon_app/package-lock.json |
Lockfile v3 / dependency resolution updates. |
heartbeat_app/package-lock.json |
Lockfile update (incl. ws version bump) and metadata adjustments. |
Review details
Files not reviewed (5)
- heartbeat_app/package-lock.json: Generated file
- rcon_app/package-lock.json: Generated file
- restart_app/package-lock.json: Generated file
- scheduler_app/package-lock.json: Generated file
- shutdown_app/package-lock.json: Generated file
- Files reviewed: 8/14 changed files
- Comments generated: 8
- Review effort level: Lite
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
Comment on lines
7
to
9
| pull_request: | ||
| branches: [ "master" ] | ||
|
|
||
| env: | ||
| # Use docker.io for Docker Hub if empty | ||
| REGISTRY: ghcr.io | ||
| # github.repository as <account>/<repo> | ||
| IMAGE_NAME: ${{ github.repository }} | ||
|
|
||
|
|
||
| jobs: |
Comment on lines
+35
to
+39
| run: | | ||
| docker build . --tag ghcr.io/compscidr/rust-server:$(date +%m-%d-%Y) | ||
| docker tag ghcr.io/compscidr/rust-server:$(date +%m-%d-%Y) ghcr.io/compscidr/rust-server:latest | ||
| docker push ghcr.io/compscidr/rust-server:$(date +%m-%d-%Y) | ||
| docker push ghcr.io/compscidr/rust-server:latest No newline at end of file |
Comment on lines
+9
to
+13
| set -o allexport | ||
| source /etc/rust/rust.env | ||
| # lets you split the seed into a separate config file to make wipes easier | ||
| source /etc/rust/seed.env | ||
| set +o allexport |
Comment on lines
+27
to
+28
| # this step requires a GH_TOKEN with read:packages write:packages and delete:packages | ||
| # tested with a classic token with the above permissions |
Comment on lines
+1
to
+4
| # nodejs-22-steamcmd-ubuntu-24.04 (Ubuntu 24.04, glibc 2.39, node 22) pinned by digest | ||
| FROM didstopia/base:nodejs-22-steamcmd-ubuntu-24.04@sha256:ae8f7bdab405f669eadcc7178627dd47def015e22227247547904b22156f5a47 | ||
|
|
||
| LABEL maintainer="Didstopia <support@didstopia.com> & Zaroxh <admin@rusttilldawn.eu>" | ||
| LABEL org.opencontainers.image.source="https://github.com/compscidr/rust-server" |
| the container will keep the same old values from when it was originally run. | ||
| This setup allows overriding anything passed in from docker. The config file | ||
| is located at `/etc/rust/rust.env` in the container. You can map a volume | ||
| locally to the file for peristence. |
| exit_handler() | ||
| { | ||
| echo "Shutdown signal received" | ||
| echo "Shutdown signal received" |
Comment on lines
+15
to
+24
| run() { | ||
| local stub | ||
| stub=$(mktemp) | ||
| printf 'echo "%s"\nexit 0\n' "$1" > "$stub" | ||
| ( | ||
| export STEAMCMD_SH=$stub STEAMCMD_RETRY_DELAY=0 | ||
| eval "$(sed -n '/^install_or_update()/,/^}/p' start_rust.sh)" | ||
| install_or_update > /dev/null 2>&1 | ||
| ) | ||
| } |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
On 2026-09-03 the weekly and monthly servers on nas.local restarted for their scheduled wipe at 18:00 UTC — the moment Facepunch's monthly depot was flipping. steamcmd printed
Error! App '258550' state is 0x486 after update job.and exited 0.start_rust.shcarried on, downloaded Oxidedevelop(2.0.7676, published 28 minutes earlier for the new build) over the old Rust install, andRustDedicatedhung at boot:A hang is not an exit, so
restart: unless-stoppednever fired, the image's own update checker could not reach RCON to restart, and both servers stayed down for five days.What
install_or_updateaccepts only steamcmd's ownSuccess! App '258550' fully installedline (printed on every boot, including already-up-to-date). Anything else sleepsSTEAMCMD_RETRY_DELAY(default 60s) and exits 1 before the Oxide step. The restart policy then retries until Steam serves the update, so a restart into a Facepunch rollout costs a few loops instead of an outage.STEAMCMD_SHandSTEAMCMD_RETRY_DELAYlettest/install_or_update_test.shdrive the extracted function with a stub steamcmd. Verified the test goes red when the guard is removed.Rollout
Ships on the next
latestbuild from main. The NAS role pulls the image, so the nextnas.ymlrun recreates both containers — do it at a quiet hour.🤖 Generated with Claude Code
https://claude.ai/code/session_01A39aALuGv7Pf3HCsbiqix8