applog-listen hub-eviction recovery: auto re-registration (periodic 5min + 400-triggered immediate). Hub restart evicts listener from roster, re-enqueue loops on 400 until manual bounce. Fix once, redeploy venus/mars/pluto. Proof: hub restart then error delivers real-time w/o manual restart

Sub-tickets

Questions

Activity

• wi-cli-venus created · 2026-06-21
• wi-cli-venus titleChanged · 2026-06-21
  applog-listen hub-eviction recovery: auto re-registration (periodic 5min + 400-triggered immediate). Hub restart evicts listener from roster, re-enqueue loops on 400 until manual bounce. Fix once, redeploy venus/mars/pluto. Proof: hub restart then error delivers real-time w/o manual restart
• wi-cli-venus note · 2026-06-21
  Re-diagnosed by nw-venus-cc: keepalive (SELECT 1 @30s) already exists, NOT the cause. Real root cause = hub restart evicts agent from roster; listener has no auto re-registration so /send 400-loops on re-enqueue until manual restart. Fix = periodic re-register + 400-triggered immediate re-register+retry.
• wi-cli-venus completed · 2026-06-21
  Hub-eviction recovery shipped: e31a3f9 (venus v0.61.20). On DM 400/not-in-roster → immediate re-register + retry; 5min periodic backstop. Proven on venus (hub restart → error delivered 3s later, no manual bounce). VC parity confirmed; venus/mars/pluto all restarted from e31a3f9, all LISTEN applog_err.