OpenBMB/ChatDev · error · RuntimeError

Failed to persist attachment '{attachment.name or attachment

Error message

Failed to persist attachment '{attachment.name or attachment.attachment_id}': {exc}

What it means

Wrapping error from _persist_message_attachments: writing a message attachment to the workspace/attachment store raised an unexpected exception (I/O error, registration failure, bad data URI). The original exception is chained via 'from exc'.

Source

Thrown at runtime/node/executor/agent_executor.py:1083

            return default_limit
        custom_limit = model.params.get("tool_loop_limit")
        if isinstance(custom_limit, int) and custom_limit > 0:
            return custom_limit
        return default_limit

    def _persist_message_attachments(self, message: Message, node_id: str) -> None:
        """Register attachments produced by model outputs to the attachment store."""
        store = self.context.global_state.get("attachment_store")
        if store is None:
            return
        for block in message.blocks():
            attachment = block.attachment
            if not attachment:
                continue
            try:
                self._persist_single_attachment(store, block, node_id)
            except Exception as exc:
                raise RuntimeError(f"Failed to persist attachment '{attachment.name or attachment.attachment_id}': {exc}") from exc

    def _persist_single_attachment(self, store: Any, block: MessageBlock, node_id: str) -> None:
        attachment = block.attachment
        if attachment is None:
            return
        if attachment.remote_file_id and not attachment.data_uri and not attachment.local_path:
            record = store.register_remote_file(
                remote_file_id=attachment.remote_file_id,
                name=attachment.name or attachment.attachment_id or "remote_file",
                mime_type=attachment.mime_type,
                size=attachment.size,
                kind=block.type,
                attachment_id=attachment.attachment_id,
            )
            block.attachment = record.ref
            return

        workspace_root = self.context.global_state.get("python_workspace_root")

View on GitHub (pinned to 4fb2db0ea9)

Solutions

1. Inspect the chained __cause__ exception — it carries the real failure
2. Verify python_workspace_root in context.global_state is writable and mounted
3. Check the attachment has valid local_path/data_uri/data_bytes before executing the node
4. Retry after fixing storage; persist attachments before generating long conversations

import os
assert os.access(workspace_root, os.W_OK), 'workspace not writable'

try:
    executor.execute(node, inputs)
except RuntimeError as e:
    if 'Failed to persist attachment' in str(e):
        cause = e.__cause__  # real storage failure
        ...

When it happens

Trigger: Any exception while persisting an attachment block: unwritable workspace directory, corrupt data_uri, store.register_file failure, permission errors during copy2.

Common situations: Read-only or full disk workspace; container missing volume mount; attachment uploaded as a data URI that fails decoding; store backend (DB/object store) outage mid-run.

Related errors

• memory store {attachment.name} not found
• Workspace or node context missing for attachment persistence
• Attachment missing data for persistence
• Failed to save workflow file
• Failed to update workflow id after rename

AI-assisted analysis of OpenBMB/ChatDev@4fb2db0ea9 (2026-08-27). Data as JSON: /api/errors/2825b67bf818355c. Report an issue: GitHub.