Fix backend's memory leak
Summary
Out of the 3 node deployment, a single node starts accumulating process memory
usage as time passes. On production this goes up by ~300MB each day.
Steps to reproduce
Run a multi-cluster deployment with heavy use. Use prometheus to measure the memory usage.
What is the result?
The memory usage grows a lot over time.
What should happen
The memory usage should resemble one of the non-affected nodes.
Edited by kenkeiras