Yes I totally agree with your thoughts of the 0 downtime redeploy issue being the ultimate issue in this case. It seems like the update in Strapi database settings always causes the Render database to restart but perhaps due to traffic or something else it gets a huge spike in connections and despite a longer waitOut time it still throws a ECONNREFUSED.
I confirm, after doing the deployment like 3-4 times in a row it suddenly works. However, this is absolutely not acceptable for me running a critical production server. I can’t deploy and hope that after 3 deployment failures it all of a sudden works… until I have to push a new update or the server restarts for an unknown reason and causes the whole problem over again.
I already reached out to Render and they told me the issue is on Strapi’s side. I feel extremely stuck and really need to update my pool size soon but there’s no way for me to do.