zed-industries/zed · error · RuntimeError

Failed to install or start judge proxy

Error message

Failed to install or start judge proxy (exit {result.return_code}){detail}

What it means

verifier.py installs and starts the judge proxy inside the target container by exec'ing a root shell script (which even tails /tmp/zed-judge-proxy.log on failure). If that exec returns a non-zero exit code, this RuntimeError is raised with the combined stdout/stderr plus the proxy log tail as detail — the install or the startup failed inside the container.

Solutions

Read the detail suffix — it includes the script output and the last lines of /tmp/zed-judge-proxy.log
Exec into the container and re-run the install script manually to reproduce
Verify egress rules allow the install step, or pre-bake the proxy into the image
Confirm the volume carrying the proxy is mounted at the path the script expects

Example fix

# before
# RuntimeError: Failed to install or start judge proxy (exit 1): ...

# after — reproduce inside the container
pier exec <job> --user-root -- /bin/sh -c 'sh -x /path/to/install-proxy.sh; tail -20 /tmp/zed-judge-proxy.log'

Defensive patterns

Strategy: try-catch

Validate before calling

result = container.exec(["python3", "--version"], user="root")
if result.return_code != 0:
    raise SystemExit("container lacks python3; pre-bake it into the image or fix the base image")

Try / catch

try:
    start_judge_proxy(container, env)
except RuntimeError as err:
    if "judge proxy" in str(err):
        log_tail = container.exec(["tail", "-20", "/tmp/zed-judge-proxy.log"], user="root").stdout
        raise RuntimeError(f"{err}\nproxy log:\n{log_tail}") from err
    raise

Prevention

Pre-bake the judge proxy into the container image instead of installing at runtime
Pin the volume mount path that carries the proxy payload
Keep an egress allowlist that covers the install step's hosts

When it happens

Trigger: The container has no egress so the install step's downloads fail; the proxy script or binary is missing from the mounted volume; python3 absent or wrong architecture in the container; the proxy starts then immediately exits, leaving a log tail in the message.

Common situations: Sandboxed eval containers with blocked networking; volume mounts misconfigured so the proxy payload path is empty; base image changes dropping python3; disk-full in /tmp.

Related errors

AI-assisted analysis of zed-industries/zed@f4178619ac (2026-08-20). Data as JSON: /api/errors/da531263a024b21b. Report an issue: GitHub.

Appendix: source

Thrown at crates/eval_cli/zed_eval/verifier.py:57

                "ensure-judge-proxy\n"
                "python3 - <<'ZED_PROXY_CHECK_EOF'\n"
                "import os, socket\n"
                "port = int(os.environ.get('ZED_JUDGE_PROXY_PORT', '8089'))\n"
                "with socket.create_connection(('127.0.0.1', port), timeout=5):\n"
                "    pass\n"
                "ZED_PROXY_CHECK_EOF\n"
                "if [ -f /tmp/zed-judge-proxy.log ]; then "
                "tail -2 /tmp/zed-judge-proxy.log; fi\n"
            ),
            user="root",
            env=env,
        )
        if result.return_code != 0:
            output = "\n".join(
                part for part in (result.stdout, result.stderr) if part
            ).strip()
            detail = f": {output}" if output else ""
            raise RuntimeError(
                f"Failed to install or start judge proxy (exit {result.return_code}){detail}"
            )

View on GitHub (pinned to f4178619ac)