sendTransaction RPC method makes the node fall out of sync

[vscorpio]

Problem

I'm running Agave 2.0.15 with Yellowstone v2.0.0 for 2.0.15.

Upon changing to Agave 2.x, I encountered an issue that kept repeating. The moment I try to send a transaction through it, using the standard RPC method, shortly after, the node falls off sync without being able to catch up anymore, no matter how long it runs. I can read and spam it as long as I want; nothing bad happens.

I'm even running a self-hosted Jupiter API through the node and can send as many Jito bundles as I want without having any issues (as it's sending the transactions externally not through my own RPC). However, the moment I make a single call to sendTransaction, the node begins slowly losing sync until the point of no return. No CPU spikes, no out of usual logs, no network spikes, no disk spikes, nothing. Just slowly losing sync.

Makes no total sense whatsoever. Anybody else had this issue?