/start entrypoint broken

[Wroud]

Describe the problem

in last update (Sep 20, 2025 at 11:58 am)

probably this not working:
9a6b858

Container definition

itzg/minecraft-server:java21-alpine

Container logs

[Paste logs here](sh: /start: not found)

i have custom startup command:

set -eu

mkdir -p /data /data-efs

echo "[data] sync data from EFS to fast storage"
if command -v rsync >/dev/null 2>&1; then
  rsync -a --delete /data-efs/ /data/ || true
else
  (cd /data-efs && tar -cf - .) | (cd /data && tar -xf -) || true
fi

export HOME=/data

cleanup() {
  echo "[data] sync data from fast storage to EFS"
  if command -v rsync >/dev/null 2>&1; then
    rsync -a --delete /data/ /data-efs/ || true
  else
    (cd /data && tar -cf - .) | (cd /data-efs && tar -xf -) || true
  fi
}

# Start server in background
/image/scripts/start &
child=$!

# Forward TERM/INT to the server, run cleanup on exit
trap "echo '[shutdown] received signal, stopping server'; kill -TERM $child; cleanup" TERM INT
trap cleanup EXIT

wait $child

[itzg]

Fixed by #3663

I highly recommend avoiding custom entry scripts for any docker image. Use init container strategy instead.

[Wroud]

I've linked the same commit, and I still had this problem.

I've already changed my setup from the legacy entry point, so it works for me now. But someone else might face the same problem.

Or maybe this fix wasn't in the last image that I had when I got this issue?