Native memory leak in Java thread‑sampling sensor causing RSS growth in Dynatrace OneAgent
0 reputation · 01 Feb 2022, 21:10 UTC
The goal is to ascertain whether the unresolved design decision around process‑group merging in Dynatrace OneAgent could leave native thread‑sampling buffers allocated after the leak fix introduced in version 1.225+.
Observed RSS growth is tied to Java thread counts and only appears in releases prior to 1.225, but the fix notes do not detail the native buffer release path or how merging decisions interact with the sensor, creating uncertainty about residual memory impact in newer agents.
Does the process‑group merging algorithm retain references to native thread‑sampling buffers after a merge operation?
Can enabling OneAgent debug‑level thread‑sampling logging show whether those buffers are freed per sampling interval in versions 1.225+?