Add a consistent pool manager
Sapessii opened this issue · comments
Describe the bug
Fix the lack of a consistent pool manager.
Basically when using supabase and loading memgpt with tons of user I get that I have reached the max number of client connections.
For reference a message on discord (https://discord.com/channels/1161736243340640419/1162177332350558339/1257022675021463654)
If you have any better alternative of the db to use please let me know!
Please describe your setup
- deployed docker on render.com
MemGPT Config
Just using the standard configuration with this db url: postgresql+pg8000://postgres.xxxxx:xxxxxx@aws-0-us-west-1.pooler.supabase.com:6543/postgres
Thank you :)
This also impacts deployments with only a couple users, 2 + the default user account.
I setup new deployment in Google Cloud Run + Supabase (Nano) and followed the steps below.
- Python client:
- Created 2 users
- Browser:
- Opened 1 browser for each (2) user account
- Python client:
- Created 2 humans
- Created 2 persona
- Created 2 agents
- Browser:
- Sent a message and then noticed the
Max client connections reached
error appear in the logs prior to receiving a response.
Note: When I built the image and ran container locally, I executed a couple 'list' commands with the python client.