fix(Postgres PGVector Store Node): Do not initialize client to avoid idle connections

[fcharlaix-opendsi]

Summary

When using Postgres PGVector Store to import a lot of documents, the Postgres server will deny connection after some items due to the number of clients :

remaining connection slots are reserved for roles with the SUPERUSER attribute
error: remaining connection slots are reserved for roles with the SUPERUSER attribute
    at /usr/local/lib/node_modules/n8n/node_modules/pg-pool/index.js:45:11
    at processTicksAndRejections (node:internal/process/task_queues:95:5)
    at PGVectorStore._initializeClient (/usr/local/lib/node_modules/n8n/node_modules/@langchain/community/dist/vectorstores/pgvector.cjs:344:23)
    at Function.initialize (/usr/local/lib/node_modules/n8n/node_modules/@langchain/community/dist/vectorstores/pgvector.cjs:336:9)
    at Function.fromDocuments (/usr/local/lib/node_modules/n8n/node_modules/@langchain/community/dist/vectorstores/pgvector.cjs:735:26)
    at Object.populateVectorStore (/usr/local/lib/node_modules/n8n/node_modules/@n8n/n8n-nodes-langchain/dist/nodes/vector_store/VectorStorePGVector/VectorStorePGVector.node.js:239:9)
    at ExecuteContext.execute (/usr/local/lib/node_modules/n8n/node_modules/@n8n/n8n-nodes-langchain/dist/nodes/vector_store/shared/createVectorStoreNode.js:206:21)
    at Workflow.runNode (/usr/local/lib/node_modules/n8n/node_modules/n8n-workflow/dist/Workflow.js:741:19)
    at /usr/local/lib/node_modules/n8n/node_modules/n8n-core/dist/WorkflowExecute.js:724:51
    at /usr/local/lib/node_modules/n8n/node_modules/n8n-core/dist/WorkflowExecute.js:1155:20

All of the Postgres clients are in idle state, waiting for the event ClientRead.

Test

Example workflow : Test_Postgres_PGVector_Store.json

1. List Postgres connected clients at reset with : SELECT count(1) FROM pg_stat_activity;
2. Run multiples time this workflow without the patch, and list again Postgres connected clients.
3. Apply the patch, rebuild, run multiples time the workflow and list again Postgres connected clients.

You should see more connection after the test without the patch, there here until the n8n server is terminated.
With the patch, you should get the same number of connections as the number at reset.

Related Linear tickets, Github issues, and Community forum posts

The fix is based of this comment of langchain-ai/langchainjs#5029.

Review / Merge checklist

• PR title and summary are descriptive. (conventions)
• Docs updated or follow-up ticket created.
• Tests included.
• PR Labeled with release/backport (if the PR is an urgent fix that needs to be backported)

[CLAassistant]

All committers have signed the CLA.

[Joffcom]

Hey @fcharlaix-opendsi,

Thanks for the PR, We have created "GHC-608" as the internal reference to get this reviewed.

One of us will be in touch if there are any changes needed, in most cases this is normally within a couple of weeks but it depends on the current workload of the team.

[fcharlaix-opendsi]

Hi, I have no idea how to properly set up a test for this bug :/

[balmacefa]

I use pg bouncer to handle this issue, of connection pool