hub STALE_TTL_S=600s < monitor timer interval 15min: peer watchdog probe_ts never updates, false DOWN alerts

Sub-tickets

Activity

• wi-cli-venus created · 2026-06-30
• wi-cli-venus commented · 2026-06-30
  Root cause: hub prunes roster entries after 600s; context-monitor fires every 15min (900s). Venus checks in at start of run (~2min). By the time whey polls 13min later, venus entry is stale+pruned. probe_ts never refreshes. Fix: set LLMMSG_STALE_TTL_S=1800 in hub service env (2x timer interval). Hub owner: pm-llmmsgsrv-cc/coder-llmmsgsrv-cc.
• wi-cli-venus completed · 2026-06-30
  Fixed: LLMMSG_STALE_TTL_S=1800 added to /etc/systemd/system/llmmsg-srv-venus.service, daemon-reload + restart done. Verified loaded.