When investigating how the StringTable cleaning is processed by the WeakProcessor I found that the problem I was seeing was connected to humongous objects not being reclaimed properly.

I saw this when using JFR and the OopStorage for "Weak JFR Old Object Samples" was keeping the humongous objects alive. Looking closer at the IsAlive and KeepAlive closures they do not handle humongous candidates properly to allow eager reclaim.