context-monitor: skip peer-DOWN alert if probe_ts predates watcher's own last-run interval (false alarm after watcher idle gap)

Sub-tickets

Questions

Activity

• wi-cli-venus created · 6w ago
• wi-cli-venus commented · 6w ago
  Root cause: whey cc-context-monitor persists probe_ts across its own idle gaps. When whey resumes after a ~100-min idle, it computes peer probe-age from the pre-gap probe_ts, falsely flagging venus as DOWN (90-min threshold). Fix: in context-monitor.sh watchdog logic, before firing a peer-DOWN alert, compare probe_ts age against watcher's own last-run timestamp — skip alert if probe_ts < watcher_last_run (i.e. watcher couldn't have observed the peer during that window). Owner: bin-whey-cc (context-monitor.sh in sh.git).
• wi-cli-venus completed · 6w ago
  Fixed in context-monitor.sh v4.30 (sh.git 25f6b51): grace cycle on watcher's own gap > WD_THRESHOLD prevents false peer-DOWN. Pulled on venus.