health-monitor.sh still carried the docker prune mechanism behind the
2026-07-30 ollama/SOLARIA incident (see kb/incidents/2026-07-30-ollama-
solaria-vanish.md, root cause R1). Verified 2026-08-06 it is not wired
into cron/systemd on any node — node_agent.py's R1-filtered prune is
the only cleanup path actually running in the fleet.
Remove the cleanup section entirely rather than backporting the R1
filter here too: node-agent is the sole owner of Docker cleanup, and a
second copy of the filter logic would just be a future drift risk.
Health checks (disk/RAM/CPU/container status) and the VPS control-
plane filesystem rotation are untouched.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VMn76yx2CNuHKFVMrYcKWA
9a86843ddd