jupyter-collaboration v3 with jupyterlab v4.5 is flooding the terminal with errors

[martinRenou]

Description

I see many

/api/collaboration/room/json:notebook:e0977c30-d773-46f5-b871-77a029000929?sessionId=fdd2b8d1-d820-4f04-a0a7-b2bdde674908', version='HTTP/1.1', remote_ip='127.0.0.1')
    Traceback (most recent call last):
      File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/tornado/web.py", line 1848, in _execute
        result = await result
                 ^^^^^^^^^^^^
      File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/jupyter_server_ydoc/handlers.py", line 220, in get
        return await super().get(*args, **kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
      File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/tornado/websocket.py", line 277, in get
        await self.ws_connection.accept_connection(self)
      File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/tornado/websocket.py", line 890, in accept_connection
        await self._accept_connection(handler)
      File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/tornado/websocket.py", line 973, in _accept_connection
        await self._receive_frame_loop()
      File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/tornado/websocket.py", line 1130, in _receive_frame_loop
        self.handler.on_ws_connection_close(self.close_code, self.close_reason)
        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
      File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/tornado/websocket.py", line 599, in on_ws_connection_close
        self.on_connection_close()
        ~~~~~~~~~~~~~~~~~~~~~~~~^^
      File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/tornado/websocket.py", line 591, in on_connection_close
        self.on_close()
        ~~~~~~~~~~~~~^^
      File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/jupyter_server_ydoc/handlers.py", line 324, in on_close
        if isinstance(self.room, DocumentRoom) and self.room.clients == [self]:
                      ^^^^^^^^^
    AttributeError: 'YDocWebSocketHandler' object has no attribute 'room'
[E 2025-11-20 10:26:19.642 YDocExtension] Jupyter Websocket Server:
      + Exception Group Traceback (most recent call last):
      |   File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/pycrdt_websocket/websocket_server.py", line 217, in start
      |     async with create_task_group() as self._task_group:
      |                ~~~~~~~~~~~~~~~~~^^
      |   File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/anyio/_backends/_asyncio.py", line 781, in __aexit__
      |     raise BaseExceptionGroup(
      |         "unhandled errors in a TaskGroup", self._exceptions
      |     ) from None
      | ExceptionGroup: unhandled errors in a TaskGroup (1 sub-exception)
      +-+---------------- 1 ----------------
        | Exception Group Traceback (most recent call last):
        |   File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/pycrdt_websocket/yroom.py", line 226, in start
        |     async with create_task_group() as self._task_group:
        |                ~~~~~~~~~~~~~~~~~^^
        |   File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/anyio/_backends/_asyncio.py", line 781, in __aexit__
        |     raise BaseExceptionGroup(
        |         "unhandled errors in a TaskGroup", self._exceptions
        |     ) from None
        | ExceptionGroup: unhandled errors in a TaskGroup (1 sub-exception)
        +-+---------------- 1 ----------------
          | Traceback (most recent call last):
          |   File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/jupyter_server_ydoc/rooms.py", line 206, in _broadcast_updates
          |     await super()._broadcast_updates()
          |   File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/pycrdt_websocket/yroom.py", line 148, in _broadcast_updates
          |     await self._task_group.start(self.ystore.start)
          |           ^^^^^^^^^^^^^^^^^^^^^^
          | AttributeError: 'NoneType' object has no attribute 'start'
          +------------------------------------
        | 
        | During handling of the above exception, another exception occurred:
        | 
        | Traceback (most recent call last):
        |   File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/pycrdt_websocket/yroom.py", line 239, in start
        |     await self.awareness.stop()
        |   File "/home/martin/micromamba/envs/test/lib/python3.14/site-packages/pycrdt/_awareness.py", line 99, in stop
        |     raise RuntimeError("Awareness not started")
        | RuntimeError: Awareness not started

Reproduce

install jupyter-collaboration 3 with jupyterlab 4.5

[davidbrochart]

It seems that with JupyterLab 4.5 the room is created with the RTC: prefix in PUT /api/collaboration/session/{path}, while it was not the case in JupyterLab 4.4.

[davidbrochart]

@jtpio mentioned that it could be caused by jupyterlab/jupyterlab#15958, and I checked that it is indeed the case.

[kuldeep-surendra]

A notebook file was automatically deleted when the Real-Time Collaboration (RTC) system crashed with a RuntimeError: Awareness not started. The file deletion appears to be triggered by an overly aggressive error recovery mechanism that incorrectly includes physical file deletion as part of its cleanup process.

Environment

jupyter-collaboration: 4.1.1
datalayer_pycrdt: 0.12.17
jupyterlab: 4.4.7
Python: 3.11
JupyterHub: (deployed in Kubernetes)

Jupyter Websocket Server:\n + Exception Group Traceback (most recent call last):\n | File \"/opt/conda/lib/python3.11/site-packages/pycrdt/websocket/websocket_server.py\", line 229, in start\n | async with create_task_group() as self._task_group:\n | File \"/opt/conda/lib/python3.11/site-packages/anyio/_backends/_asyncio.py\", line 781, in __aexit__\n | raise BaseExceptionGroup(\n | ExceptionGroup: unhandled errors in a TaskGroup (1 sub-exception)\n +-+---------------- 1 ----------------\n | Exception Group Traceback (most recent call last):\n | File \"/opt/conda/lib/python3.11/site-packages/pycrdt/websocket/yroom.py\", line 234, in start\n | async with create_task_group() as self._task_group:\n | File \"/opt/conda/lib/python3.11/site-packages/anyio/_backends/_asyncio.py\", line 781, in __aexit__\n | raise BaseExceptionGroup(\n | ExceptionGroup: unhandled errors in a TaskGroup (2 sub-exceptions)\n +-+---------------- 1 ----------------\n | Traceback (most recent call last):\n | File \"/opt/conda/lib/python3.11/site-packages/pycrdt/websocket/yroom.py\", line 235, in start\n | await self._start(task_status)\n | File \"/opt/conda/lib/python3.11/site-packages/pycrdt/websocket/yroom.py\", line 258, in _start\n | self._task_group.start_soon(self._run_provider)\n | ^^^^^^^^^^^^^^^^^^^^^^^^^^^\n | AttributeError: 'NoneType' object has no attribute 'start_soon'\n +---------------- 2 ----------------\n | Traceback (most recent call last):\n | File \"/opt/conda/lib/python3.11/site-packages/jupyter_server_ydoc/rooms.py\", line 206, in _broadcast_updates\n | await super()._broadcast_updates()\n | File \"/opt/conda/lib/python3.11/site-packages/pycrdt/websocket/yroom.py\", line 165, in _broadcast_updates\n | await self._task_group.start(self.ystore.start)\n | ^^^^^^^^^^^^^^^^^^^^^^\n | AttributeError: 'NoneType' object has no attribute 'start'\n +------------------------------------\n | \n | During handling of the above exception, another exception occurred:\n | \n | Traceback (most recent call last):\n | File \"/opt/conda/lib/python3.11/site-packages/pycrdt/websocket/yroom.py\", line 238, in start\n | await self.awareness.stop()\n | File \"/opt/conda/lib/python3.11/site-packages/pycrdt/_awareness.py\", line 99, in stop\n | raise RuntimeError(\"Awareness not started\")\n | RuntimeError: Awareness not started",

[krassowski]

The file deletion appears to be triggered by an overly aggressive error recovery mechanism that incorrectly includes physical file deletion as part of its cleanup process

That would be pretty bad, can you point to a line of code that you believe is causing this?

Also, just to spell it out, this can be worked around by upgrading to jupyter-collaboration v4

[martinRenou]

it seems @kuldeep-surendra is saying they're using jupyter-collaboration: 4.1.1 with JupyterLab 4.4.7.

[martinRenou]

Also, just to spell it out, this can be worked around by upgrading to jupyter-collaboration v4

We discovered the issue with JupyterGIS which is currently stuck with jupyter-collaboration v3 until we sort out the ContentProvider concerns for lite (we don't want to make JupyterGIS incompatible with lite)

[kuldeep-surendra]

@krassowski I am using jupyter-collaboration v4
I cannot point to the line of code that can be causing this, But here is the sequence of events from the logs that i observed.

1. Log entry [I 2025-11-18 02:18:29.532 ServerApp] Room json:notebook:1e03ec94-5188-4ff6-a78c-89d3a3349b5a deleted
2. Followed by: [I 2025-11-18 02:18:29.585 ServerApp] Deleting file notebook.ipynb

This shows that room deletion triggered file deletion, which is the bug.

• Room cleanup happens at 02:18:29.532
• File deletion happens at 02:18:29.585 (53ms later)

[davidbrochart]

I think that the logs are misleading, Deleting file notebook.ipynb doesn't actually means the file is deleted, but that an entry is removed in FileLoaderMapping.

[martinRenou]

Has anyone been able to find a potential fix for this?

[jtpio]

it could be caused by jupyterlab/jupyterlab#15958, and I checked that it is indeed the case.

How did you check?

The same PR seems to be causing issues with other custom drives, as noticed in jupyterlite/jupyterlite#1794. From the investigation in jupyterlite/jupyterlite#1794 it was mostly related to the handling of files at the top-level. So not sure this is exactly the issue as the one reported here.

[davidbrochart]

it could be caused by jupyterlab/jupyterlab#15958, and I checked that it is indeed the case.

How did you check?

I reverted that PR locally and didn't see any error anymore.