can1357/oh-my-pi · error · GitCommandError

GitCommandError: git checkout --detach failed (nonzero exit)

Error message

GitCommandError: git checkout --detach failed (nonzero exit)

What it means

ensure_release_workspace() runs `git checkout --detach` in the shared clone pool so the release worktree can be cut from the default-branch commit. A nonzero exit raises GitCommandError, aborting release workspace setup before `git worktree add` runs.

Source

Thrown at python/robomp/src/sandbox.py:1056

            pool = self.ensure_clone(
                repo=repo,
                clone_url=clone_url,
                default_branch=default_branch,
                refresh=False,
            )
            self.transport.fetch_pool(repo=repo, pool_dir=pool)
            ws_root = self.workspace_root(repo, "release")
            repo_dir = ws_root / "repo"
            session_dir = ws_root / f".omp-session-{tag}"
            context_dir = ws_root / "context"
            artifacts_dir = ws_root / "artifacts"
            for path in (ws_root, session_dir, context_dir, context_dir / "repro", artifacts_dir):
                path.mkdir(parents=True, exist_ok=True)

            detach = ["git", "checkout", "--detach"]
            detached = _safe_run(detach, cwd=pool)
            if detached.returncode != 0:
                raise GitCommandError(detach, detached.returncode, detached.stdout, detached.stderr)

            if not (repo_dir / ".git").exists():
                _worktree_add(
                    [
                        "git",
                        "worktree",
                        "add",
                        "-B",
                        default_branch,
                        str(repo_dir),
                        f"origin/{default_branch}",
                    ],
                    pool=pool,
                    repo_dir=repo_dir,
                )

            _share_git_metadata_with_slots(repo_dir, slot_uid)
            _provision_runtime_dirs(ws_root)

View on GitHub (pinned to 9690622007)

Solutions

1. Inspect GitCommandError.stderr — 'index.lock exists' means remove the stale lock, 'invalid object' means the pool clone is corrupt.
2. Delete the affected pool clone (under /data) and let the retry re-clone from the remote.
3. Verify the upstream repo actually has a default branch with commits.
4. If caused by concurrency, avoid running a second orchestrator against the same /data volume.

proc = subprocess.run(['git', 'rev-parse', 'HEAD'], cwd=pool, capture_output=True)
if proc.returncode != 0:
    raise RuntimeError(f'pool clone unusable: {proc.stderr.decode()}')

from robomp.sandbox import GitCommandError

def has_returncode(exc: BaseException, code: int) -> bool:
    return isinstance(exc, GitCommandError) and exc.returncode == code

try:
    ws = manager.ensure_release_workspace(repo=repo, tag=tag)
except GitCommandError as e:
    if e.returncode == 124:
        ...  # timeout: check for hung git / locks before retrying
    else:
        ...  # re-clone pool and retry once

When it happens

Trigger: Calling ensure_release_workspace() when the pool clone has no commits (empty upstream repo), is mid-rebase/merge with conflicts, has a locked index, or `git checkout --detach` hits a corrupt object database.

Common situations: First release event for a repo whose clone pool was seeded from an empty or unreachable remote; concurrent process holding the pool's index.lock; corrupted clone after a container kill.

Related errors

• GitCommandError: git command failed (nonzero exit)
• Isolated subagent execution could not be prepared: ${message
• could not find an unused worktree path under ${basePath} (tr
• git timed out after {effective_timeout:.0f}s: {' '.join(_red
• GitCommandError(cmd, proc.returncode, proc.stdout, proc.stde

AI-assisted analysis of can1357/oh-my-pi@9690622007 (2026-08-31). Data as JSON: /api/errors/c6f782c1f2b4ad85. Report an issue: GitHub.