≥300ms of GC isn't ideal, usually a sign of pushing a cluster too hard for its size, although it's not too bad if it's only happening a couple of times a minute.
This is also a DEBUG-level message which the manual says is "only intended for expert use".
I don't think posting small fragments of randomly-selected log messages and metrics is a very good way to get useful help in a forum like this. If you're having a specific problem then you would do better to spend some time describing all of its symptoms and sharing as full a picture as possible in a single post. I suggest a new thread for that too, this one is no longer about your claim of a "strange memory leak" as the title suggests.