Make broker-unavailable failures actionable #1354

Merged
coilyco-ops merged 1 commit from issue-1342 into main 2026-07-15 08:33:26 +00:00
Member

This change makes broker contact failures actionable across director surfaces. Read-only brokered commands now report a typed broker diagnostic instead of falling back to local launch/list behavior, and the diagnostics name the broker address source, connection failure, and a single remediation step.

closes #1342
ward.workflow: pull-request-and-merge

  • Added broker-unreachable and broker-timeout diagnostics, then wired them through ref dispatch, list/ps, stop, PR workflow, and freeform engineer forwarding.
  • Added refused-connection and timeout fixtures, plus a capacity-defer wording check so capacity backpressure stays distinct from broker failures.
  • Live host broker health was not verified from the sealed engineer container; verification was limited to unit fixtures and broker-response tests.
This change makes broker contact failures actionable across director surfaces. Read-only brokered commands now report a typed broker diagnostic instead of falling back to local launch/list behavior, and the diagnostics name the broker address source, connection failure, and a single remediation step. closes #1342 ward.workflow: pull-request-and-merge - Added `broker-unreachable` and `broker-timeout` diagnostics, then wired them through ref dispatch, list/ps, stop, PR workflow, and freeform engineer forwarding. - Added refused-connection and timeout fixtures, plus a capacity-defer wording check so capacity backpressure stays distinct from broker failures. - Live host broker health was not verified from the sealed engineer container; verification was limited to unit fixtures and broker-response tests.
Make broker-unavailable failures actionable
All checks were successful
test / test (pull_request) Successful in 2m12s
c4e2dd43bb
Sign in to join this conversation.
No description provided.