Hi, I have multiple services in my project (workspace "zonal-transformation", environment "production": stratos — a Python worker, stratos-backend — FastAPI/Uvicorn, stratos-frontend — Streamlit) that keep restarting repeatedly with no obvious cause on the application side. I've already ruled out several things before writing in:

• Not memory: checked the Metrics tab several times during the restarts — max memory observed was ~250MB, no sign of OOM (no "flat ceiling then drop to zero" shape typical of an OOM kill; the graph stays irregular, consistent with normal usage).
• Not an overlapping redeploy: checked the timestamps of my GitHub pushes — no push happened in the windows where the services restarted.
• Not an unhandled Python exception: added PYTHONUNBUFFERED=1 on the frontend to make sure a traceback wasn't just sitting in a stdout buffer — still zero tracebacks in the logs. The logs only show a clean shutdown sequence: "Stopping..." → "Stopping Container" → "Starting Container" (frontend, Streamlit) and "Shutting down" → "Waiting for application shutdown" → "Application shutdown complete" → "Finished server process [1]" (backend, Uvicorn) — a normal SIGTERM-driven graceful shutdown, not a crash with a traceback.
• Manually reviewed the backend's write-endpoint code (POST/DELETE on /posizioni and /alert, the ones exercised when the issue shows up) — standard pattern (open connection, query, commit, close in a finally), no plausible unhandled exception, nothing that looks like a code bug here.
• Not tied exclusively to one specific user action: it initially looked like it only fired when clicking "Delete" in a Streamlit form (st.rerun()), but it has also happened with nobody interacting with the app at all.
• Strongest clue: in different occurrences, stratos-frontend (Streamlit), stratos-backend (FastAPI/Uvicorn), and stratos (a third, unrelated Python scheduler process) have all restarted — three independent tech stacks with no shared code. It seems unlikely to be an isolated app bug repeating itself in all three; it points more toward something at the environment/project level.
• The Deployments tab doesn't show an explicit reason (e.g. "Healthcheck failed") for these instances — just the stop/start cycle.

Observed frequency: several restarts on the evening of 07/10 (~22:45-22:56) and again this morning on 07/11 (two episodes, ~06:42-06:50 and ~06:56), at irregular intervals ranging from 20 seconds to a few minutes apart.

Could you check on your end (platform logs, node scheduling/migration events, or a default healthcheck I'm not seeing) what's causing these restarts in the "production" environment of this project?

Thanks