WTH Isn't it easier to find a memory leak?

Would be nice to have a ‘simpler’ method to track down a memory leak. Would be awesome to have a way for HA to more simply identify the culprit.

There is, the mem profiler integration.

The problem is, reading and understanding where and what creates memory requires a person who intimately understands how HA works under the hood. Not even an AI (in their current state) understands that, they would likely be able to identify the memory but not the source.

Right and agreed. Which was the reason for asking for an easier way… I’m not that guy who understands it enough and hate bugging those that do. Was thinking someone smarter than I could come up w/ some kind of algo to parse / identify necessary data to identify the culprit.

I’ve had to help debugging a few times, it’s not easy. I don’t know how it could be automated without just showing graphs. But there would be like 8000 graphs.

it would be helpful to provide sensors with memory usage of each component.
I’m not sure it’s doable in current architecture of HA.

Anyway, the first step anyone can do is monitoring memory allocated by each docker. It gives overview on HA versus add-ons memory usage. I had a situation that the add-on was responsible for consuming all memory