Investigate fresh setup failures and rejected OpenAI chat turns #1652

Closed
opened 2026-10-27 13:41:15 +00:00 by sam · 0 comments
Owner

After a fresh setup, a Telegram greeting reaches the gateway but fails with OpenAI HTTP 400: function tools with reasoning_effort are unsupported for the selected model on Chat Completions. Earlier startup logs also show unavailable secret engines, an incompletely configured email channel, and update artifact failures. Trace setup output and current runtime state, distinguish historical failures from active defects, and identify the smallest corrections. Acceptance: each observed failure has an evidenced cause and a concrete correction or separately tracked follow-up.

Notes

Live investigation: all five services are active and report version 1.51.0. The reported greeting reaches the gateway and fails with OpenAI HTTP 400 for tool use with default reasoning on Chat Completions. Current tree already pins ai-sdk v0.1.43 (ee5d6f78), which selects Responses for GPT-6 tool requests. Release v1.51.1 artifact was published about six minutes after the installed services last started; deployment is still on the preceding release. Earlier missing-secret-engine failures ceased after the last restart. Email extension failed at that restart because required settings were absent; configure it or disable it. Earlier update attempts failed looking for the removed update-check artifact; current installer no longer lists it. Readiness checks validate config, disk and service contracts but do not exercise a tool-enabled model turn. No deployment or configuration changes made. Next correction: install the existing v1.51.1 release and verify a real chat turn, then resolve the email extension configuration.

Close reason

Stale: fixed days ago; reopened by a bd github sync


Imported from beads archie-core-wrv2 · GitHub #1523 · type bug · priority P1 · created 2026-10-08 · closed 2026-10-10

After a fresh setup, a Telegram greeting reaches the gateway but fails with OpenAI HTTP 400: function tools with reasoning_effort are unsupported for the selected model on Chat Completions. Earlier startup logs also show unavailable secret engines, an incompletely configured email channel, and update artifact failures. Trace setup output and current runtime state, distinguish historical failures from active defects, and identify the smallest corrections. Acceptance: each observed failure has an evidenced cause and a concrete correction or separately tracked follow-up. ### Notes Live investigation: all five services are active and report version 1.51.0. The reported greeting reaches the gateway and fails with OpenAI HTTP 400 for tool use with default reasoning on Chat Completions. Current tree already pins ai-sdk v0.1.43 (ee5d6f78), which selects Responses for GPT-6 tool requests. Release v1.51.1 artifact was published about six minutes after the installed services last started; deployment is still on the preceding release. Earlier missing-secret-engine failures ceased after the last restart. Email extension failed at that restart because required settings were absent; configure it or disable it. Earlier update attempts failed looking for the removed update-check artifact; current installer no longer lists it. Readiness checks validate config, disk and service contracts but do not exercise a tool-enabled model turn. No deployment or configuration changes made. Next correction: install the existing v1.51.1 release and verify a real chat turn, then resolve the email extension configuration. ### Close reason Stale: fixed days ago; reopened by a bd github sync --- _Imported from beads `archie-core-wrv2` · GitHub #1523 · type bug · priority P1 · created 2026-10-08 · closed 2026-10-10_
sam added the highbug labels 2026-10-27 13:41:15 +00:00
sam self-assigned this 2026-10-27 13:41:15 +00:00
sam closed this issue 2026-10-27 13:41:15 +00:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: sam/archie#1652