table_catalog error when deploying on AWS Lambda

[dvd3v]

I'm trying to run TiMVT in AWS Lambda + Aurora (based on https://github.com/developmentseed/serverless-timvt). This is how my application looks like:

from timvt.db import close_db_connection, connect_to_db, register_table_catalog
from timvt.factory import VectorTilerFactory
from timvt.layer import FunctionRegistry
from fastapi import FastAPI, Request

# Create Application.
app = FastAPI()

# Add Function registry to the application state
app.state.timvt_function_catalog = FunctionRegistry()

# Register Start/Stop application event handler to setup/stop the database connection
# and populate `app.state.table_catalog`
@app.on_event("startup")
async def startup_event():
    """Application startup: register the database connection and create table list."""
    await connect_to_db(app)
    await register_table_catalog(
        app
    )

@app.on_event("shutdown")
async def shutdown_event():
    """Application shutdown: de-register the database connection."""
    await close_db_connection(app)

# Register endpoints.
mvt_tiler = VectorTilerFactory(
    with_tables_metadata=True,
    with_functions_metadata=True,  # add Functions metadata endpoints (/functions.json, /{function_name}.json)
    with_viewer=True,
)
app.include_router(mvt_tiler.router, tags=["Tiles"])

Note; i've added the register_table_catalog function so it loads the tables from the db. Everything works fine locally, also using the Aurora database. However when I deploy the application to AWS Lambda I run into a few problems. I'm using the geolambda layers from https://github.com/lambgeo/geo-layer.

When using Python 3.8 with GDAL 3.2, all the dependencies load correctly, and I'm able to run the app. However when I call the /tables.json endpoint I run into the following error:

[ERROR] 2022-10-12T13:15:46.051Z    9e67c286-5679-44cd-b502-b709978cea4b    An error occurred running the application.
Traceback (most recent call last):
  File "/var/task/starlette/datastructures.py", line 699, in __getattr__
    return self._state[key]
KeyError: 'table_catalog'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/var/task/mangum/protocols/http.py", line 58, in run
    await app(self.scope, self.receive, self.send)
  File "/var/task/fastapi/applications.py", line 270, in __call__
    await super().__call__(scope, receive, send)
  File "/var/task/starlette/applications.py", line 124, in __call__
    await self.middleware_stack(scope, receive, send)
  File "/var/task/starlette/middleware/errors.py", line 184, in __call__
    raise exc
  File "/var/task/starlette/middleware/errors.py", line 162, in __call__
    await self.app(scope, receive, _send)
  File "/var/task/starlette/middleware/exceptions.py", line 75, in __call__
    raise exc
  File "/var/task/starlette/middleware/exceptions.py", line 64, in __call__
    await self.app(scope, receive, sender)
  File "/var/task/fastapi/middleware/asyncexitstack.py", line 21, in __call__
    raise e
  File "/var/task/fastapi/middleware/asyncexitstack.py", line 18, in __call__
    await self.app(scope, receive, send)
  File "/var/task/starlette/routing.py", line 680, in __call__
    await route.handle(scope, receive, send)
  File "/var/task/starlette/routing.py", line 275, in handle
    await self.app(scope, receive, send)
  File "/var/task/starlette/routing.py", line 65, in app
    response = await func(request)
  File "/var/task/fastapi/routing.py", line 231, in app
    raw_response = await run_endpoint_function(
  File "/var/task/fastapi/routing.py", line 160, in run_endpoint_function
    return await dependant.call(**values)
  File "/var/task/timvt/factory.py", line 232, in tables_index
    for r in request.app.state.table_catalog
  File "/var/task/starlette/datastructures.py", line 702, in __getattr__
    raise AttributeError(message.format(self.__class__.__name__, key))
AttributeError: 'State' object has no attribute 'table_catalog'

When using a Python 3.7 or 3.9 environment, the application does not load and returns the following error:

[ERROR] Runtime.ImportModuleError: Unable to import module 'handler': No module named 'asyncpg.protocol.protocol'
Traceback (most recent call last):

This could be related to the c-bindings of the asyncpg package.

I cannot reproduce these errors locally, so it has something to do with the AWS Lambda runtime environment. Any ideas?

[vincentsarago]

@dvd3v do you have a link to the code you are using?

This could be a issue with your handler. If it's not the case try setting lifespan="auto" see: https://github.com/developmentseed/eoAPI/blob/master/stack/handlers/vector_handler.py#L12

When using a Python 3.7 or 3.9 environment, the application does not load and returns the following error:

This is a dependency issue, and could be because of a lot of things.

[dvd3v]

Here's the code I'm using: https://github.com/dvd3v/serverless_timvt (note that the Aurora DB is not in here as it was setup in a different project.)

When setting the lifespan="auto", I get timeout errors. Setting it to off works and them I'm able to call the endpoints.

Regarding the table_catalog error, I've noticed this only happens when installing TiMVT via pip (pip install timvt). When installing from source (pip install git+https://github.com/developmentseed/timvt.git#egg=timvt), the error does not occur. So it's related to timvt 0.7.0. However, the /tables.json endpoint then returns emtpy values:

Expected:

[
  {
    "id": "public.sentinel_mgrs",
    "bounds": [
      -180,
      -83.83595275878906,
      180,
      83.74835205078125
    ],
    "minzoom": 0,
    "maxzoom": 22,
    "default_tms": "WebMercatorQuad",
    "tileurl": "http://127.0.0.1:8003/tiles/public.sentinel_mgrs/{z}/{x}/{y}",
    "type": "Table",
    "schema": "public",
    "table": "sentinel_mgrs",
    "geometry_type": "POLYGON",
    "geometry_column": "geom",
    "geometry_srid": 4326,
    "properties": {
      "id": "varchar",
      "geom": "geometry",
      "name": "varchar",
      "ogc_fid": "int4"
    }
  }
]

Return from Lambda:

[]

The Lambda has access to the DB and locally it returns the correct data.

[vincentsarago]

there are a lot of change that haven't been published yet https://github.com/developmentseed/timvt/blob/master/CHANGES.md#next-tbd