[BUG] Unable to save timeseries document: lock for timeseries is not allowed

[tyrantlink]

Describe the bug
Document.save() fails when using a timeseries collection, returning pymongo.errors.OperationFailure: Taking test.timeseries_test lock for timeseries is not allowed, full error:

{'ok': 0.0, 'errmsg': 'Taking test.timeseries_test lock for timeseries is not allowed', 'code': 166, 'codeName': 'CommandNotSupportedOnView'}

To Reproduce
with beanie==1.26.0, pymongo==4.8.0, and motor==3.5.1 on python 3.11.5

from datetime import datetime
from asyncio import run

from beanie import init_beanie, Document, TimeSeriesConfig
from motor.motor_asyncio import AsyncIOMotorClient
from pydantic import Field


MONGO_URI = ''
DATABASE = ''


class TimeSeriesTest(Document):
    ts: datetime = Field(default_factory=datetime.now)

    class Settings:
        name = 'timeseries_test'
        timeseries = TimeSeriesConfig(
            time_field='ts',
            expire_after_seconds=60
        )


async def main():
    _client = AsyncIOMotorClient(
        MONGO_URI,
        serverSelectionTimeoutMS=5000
    )[DATABASE]

    await init_beanie(_client, document_models=[TimeSeriesTest])

    ts_test = TimeSeriesTest()

    await ts_test.save()


if __name__ == '__main__':
    run(main())

Expected behavior
Document is successfully inserted

Additional context
The code above does successfully insert when not using a timeseries collection

[szyg]

Am having the same issue. Is there any recommended work around other than just converting to a non time series collection?

[tyrantlink]

For using a time series collection, I wasn't able to find one, however for my specific use-case, just a regular collection with a TTL index fit my needs, although it's less efficient.

[baum314nk]

You must not use the save method as it passes upsert=True to the underlying pymongo function which is not supported (see here). Instead use insert. I think this should be added in the docs related to time series