Repository navigation
fix(hosts): bound reconnect churn while pending activation keeps failing #106
Description
While a stored pending certificate's activation keeps failing, the agent reconnects roughly every 31 s (the renewal retry bound from #91), for hours or days if the active certificate is far from expiry. That churns liveness and connection replacement on the control plane. It is not a security or availability failure.
Acceptance criteria
- Decide a retry policy for repeated activation failures (for example retry only inside the 6 h renewal window, or back off the interval after repeated failures), and record it in
specs/006-whr-cli/contracts/host-enrollment.md. - Implement it with tests;
make checkis green.
The other nits originally listed here are fixed directly on the pilot stack (owner decision H65:b, superseding H65:a).
🤖 Generated with Claude Code
Activity
[-]chore: collected non-blocking review nits from the pilot stack[/-][+]fix(hosts): bound reconnect churn while pending activation keeps failing[/+]test(hosts): expose activation retry on early reconnect
fix(hosts): bound repeated pending activation retries
docs(hosts): record pending activation retry evidence
test(hosts): expose wall-clock retry window transition
fix(hosts): honor UTC retry deadlines across clock changes