We recently had some 500's from nginx itself that somehow were not logged (we have screenshots, but nothing in the logs). That is weird in itself, because usually errors show up there. Regardless, I am wondering if there is something like a connection pool size that if maxed out would result in a 500? We have correlated it potentially to a recent spike in traffic, but it is not conclusive.
Anyone have any ideas of how to begin to approach such an issue?