Knex timeout - what is the best approach to fix it?

[vividcode]

I have my backend hosted at Amazon Linux AMI I made some 6 months ago. Much before that, I was using Node 17.x, which I used without much thinking.

As I kept observing - Every DB query invocation (SELECTs included) after a long time resulted in the knex time out error:

Knex: Timeout acquiring a connection. The pool is probably full. Are you missing a .transacting(trx) call?

Now this is a well-known problem and people have tried fixing it either fixing Objection / Knex version or by tweaking their node version.

I tried the former, but did not succeed. I nontheless upgraded to:

"knex": "^3.1.0"
"objection": "^3.1.4"

Finally, I succeeded by downgrading to Node v12.22.12 (Erbium) through NVM.
The process was not without its fair share of warnings emanating from Objection and Knex.

I don't see any problems with this older Node version so far.

However, I am worried if this is a good approach in the long term, and if there is any knex / Objection config tweak I am discarding without trying.

My prod config is simple (nothing well thought-out, suggestions openly welcome!)

client: "postgresql",
    connection: {
      host: "<host>",
      database: "<dbname>",
      user: "<user>",
      password: "<password>",
    },
    pool: {
      min: 2,
      max: 10,
    },
    migrations: {
      tableName: "knex_migrations",
    },

In a nutshell:

Should I make a new Amazon VM Image with Node 12.2 to make it stick, or should I try any knex config changes to see if they work with latest Node version?

Note: AWS Linux that I use currently prevents me from upgrading Node through NVM beyond 17.x for some reason - so upgrading Node isn't an option - unless I am missing any obvious solution there.

I am going live with this in 5 days, so any help will be highly appreciated!