Skip to content

Upstream Source

This page is part of Truthound Orchestration 3.x.

Source repository: seadonggyun4/truthound-orchestration Upstream docs path: docs/adoption-playbook.md Edit upstream page: Edit in orchestration

Platform Adoption Playbook

This playbook is the operator-facing bridge between "the adapter works" and "the adapter is now a supported part of the platform". It is intentionally cross-platform and sits above the host-specific pages.

Who This Is For

  • platform teams adopting Truthound across more than one orchestration host
  • operators standardizing rollout order and ownership boundaries
  • engineering leads choosing how to phase adapter adoption by workload type

When To Use It

Use this page when:

  • one team has succeeded locally and now others want the same pattern
  • multiple hosts such as Airflow, Prefect, and dbt are in scope at once
  • you need a rollout plan that stays aligned with support matrix and CI coverage

Prerequisites

  • a supported Truthound 3.x release
  • at least one validated host/platform pairing from the compatibility matrix
  • a documented owner for rules, secrets, notifications, and incident response

Minimal Quickstart

Recommended adoption order:

  1. validate one representative dataset on the chosen host
  2. document the source shape, rule set, and expected result contract
  3. add CI coverage for the host-specific path
  4. wire alerting and operational ownership
  5. roll the pattern out to adjacent datasets

Production Pattern

Use this decision table across the adapter line:

Question Recommended Decision
where should quality run? pick the host that already owns orchestration for that workload
when is zero-config enough? only for first runs and clearly local sources
when should secrets become explicit? before production SQL or remote-source execution
what is the release gate? supported tuple + preflight pass + documented result consumer

Failure Modes and Troubleshooting

Symptom Likely Cause What To Do
each team adopts a different pattern no shared playbook exists publish one blessed host-specific path per workload type
quality checks exist but no one owns incidents rollout focused on code, not operations define owner and escalation before general adoption
support matrix is ignored during rollout adoption happened from examples only make compatibility review part of promotion criteria