Connecting Self-Hosted MongoDB to AgGrid Table in Windmill UI - Need Support with Data Display Issue
Hi!
I'm trying to connect an AgGrid Table to a self-hosted MongoDB instance. Since Windmill docs currently only cover MongoDB Atlas, I'm using a custom Python script as a data source for the AgGrid.
Current Setup:
- Self-hosted MongoDB with collection containing ~100k records
- Python script using pymongo to fetch and return data
- AgGrid Table component with infinite scroll configuration
Issue:
The script executes successfully and returns properly formatted data (verified via logs), but the AgGrid Table remains empty. I've tried:
- Verifying column definitions match the data structure
- Setting Row Id Col to "unique_ID"
- Toggling pagination settings
- Adjusting infinite scroll configuration
- Using both the columnDefs in the script return and in the AgGrid configuration
The script is successfully connecting to MongoDB and returning data (verified through debug logging), but the AgGrid Table doesn't display any of it.
Is there a simpler way to connect AgGrid directly to a self-hosted MongoDB? Or if not, what am I missing in my current workaround setup?
Thanks for any help!
6 Replies
Script Structure if more context is needed:
import wmill
import pymongo
from typing import Dict
def main(grid_params: Dict = None):
# Get MongoDB connection variables
MONGO_USERNAME = wmill.get_variable("u/pruveadmin/MONGO_USERNAME")
MONGO_PASSWORD = wmill.get_variable("u/pruveadmin/MONGO_PASSWORD")
MONGO_HOST = wmill.get_variable("u/pruveadmin/MONGO_HOST")
MONGO_PORT = wmill.get_variable("u/pruveadmin/MONGO_PORT")
DB_NAME = {redacted for privacy}
mongo_uri = f"mongodb://{MONGO_USERNAME}:{MONGO_PASSWORD}@{MONGO_HOST}:{MONGO_PORT}"
# Example showing structure with unique_ID, other fields removed for privacy
projection = {
"_id": 0,
"unique_ID": 1,
# ... other fields
}
try:
client = pymongo.MongoClient(mongo_uri)
db = client[DB_NAME]
collection = db.CD_proc
cursor = collection.find({}, projection).limit(100)
rows = list(cursor)
return {
"columnDefs": [
{"field": "unique_ID", "headerName": "ID", "width": 120},
# ... other columns
],
"rowData": rows,
"lastRow": -1,
"pagination": True,
"rowModelType": "infinite",
"cacheBlockSize": 100,
"paginationPageSize": 100
}
except Exception as e:
return {"error": str(e)}
finally:
if 'client' in locals():
client.close()
AgGrid Configuration:
{
"rowModelType": "infinite",
"cacheBlockSize": 100,
"paginationPageSize": 100,
"rowBuffer": 0,
"getRowId": "params => params.data.unique_ID",
"infiniteInitialRowCount": 1,
"maxBlocksInCache": 2,
"maxConcurrentDatasourceRequests": 2,
"serverSideInfiniteScroll": true,
"pagination": true,
"defaultColDef": {
"filter": true,
"sortable": true,
"resizable": true,
"floatingFilter": true
}
}
@rubenf any thoughts or insights?
I would start with a simpler example and isolate the issue step by step
Can you anticipate when self-hosted MongoDB is part of standard support?
when we have enough EE customers needing it
@AntPruve my 5c It would be easier to read/reconstruct later if you return an object pre-built with data rather than build it when returning (i.e. response = {"columndefs etc"}; return response}
also try mocking a working response before transforming into one so you can eliminate the point of failure in this flow.
thanks for your 5c - it's appreciated!
I agree. I will circle back and try this