How can I fix the 'Cannot use a pool after calling end on the pool' error in my AI Agent using Postgres Chat Memory with Supabase?

The error message:

Cannot use a pool after calling end on the pool

indicates that the connection pool—which manages connections to your Supabase (PostgreSQL) database—has already been terminated (via the end() call) and is now being reused. This typically happens when the node or underlying library inadvertently closes the pool, while a part of your workflow is still trying to interact with it.

Troubleshooting Steps

1. Review Workflow Nodes

   • Verify that the Postgres Chat Memory node isn’t being reinitialized or duplicated in a way that calls for closing the pool early.
   • Consider removing the Postgres Chat Memory node from your workflow and re-adding it to ensure you’re using the most recent version of the node’s configuration.

2. Check Node and Library Versions

   • Since you’re running n8n version 1.74.1 (Self Hosted), confirm whether there have been any updates or known issues with the Postgres Chat Memory node regarding connection management, especially with Supabase.

3. Connection Pool Management

   • Look into your node configuration and any custom parameters to ensure that the database connection isn’t being explicitly ended. If you have any additional scripts or custom logic that interact with the connection pool, confirm they aren’t calling end() prematurely.

4. Database Client Behavior

   • Since Supabase uses PostgreSQL behind the scenes, review any logs or behavior from the database client library. The error is common when a client tries to run queries on a pool that has already been closed.

Following these steps should help you pinpoint why the pool is being closed and allow you to adjust the configuration in the Postgres Chat Memory node or your Supabase connection settings accordingly.