NestJS Ram Spikes

Every few weeks, we get absolutely massive spikes in RAM on our servers and our response times slow to a crawl. We are running 5 replicas, so I'm not sure why this is not spreading out across different servers. I'm sure this is a problem somewhere within my code or design... I'm logging execution times on methods via an HTTP request to betterstack (the only way I've found to get logging data out of the railway servers), but I think I need a new strategy, as we can't have downtime like this, and looking at the data, I don't see anything that is too abnormal or strange. How do I create a more accurate pinpoint of this issue, or figure out more analytics surrounding why the RAM usage is so high? I know this might not be a strictly Railway problem, but there's lots of fantastic engineers here so I'd like to know if you guys have any suggestions. I feel like most of our code is pretty standard stuff, with nothing except data base insertions, some queries, etc.