Hook Managing Domo dataflow schedules across dozens of instances with ad-hoc scripts works — until you wake up to unexpected configuration drift and no idea which of your 22 daily scripts caused it.
Why It Matters Running reconciliation scripts unconditionally every day creates noisy activity logs and, worse, obscures the root cause when something changes unexpectedly. Adopting Infrastructure as Code patterns for Domo gives you a single source of truth for desired state, reviewable change PRs before anything touches production, and an audit trail that makes debugging drift tractable. It also eliminates the "who wrote that script?" problem entirely by collapsing scattered imperative scripts into a declarative manifest checked into version control.
What You'll Learn
- Declare Domo dataset schedules as manifest files that represent desired state
- Build a reconciler that compares declared state against live Domo API configuration
- Generate reviewable GitHub Pull Requests for any detected drift — before changes apply
- Set up a GitHub Actions workflow for scheduled, automated reconciliation
- Implement idempotent apply operations gated behind explicit PR approval
From Noisy Scripts to Declarative Infrastructure
The core insight is borrowed directly from Terraform: instead of writing imperative "do this" scripts that fire every day regardless of whether anything needs to change, you declare what you want and let a reconciler compute what's different.
The manifest file is the foundation — a structured declaration of which Domo datasets should run on which schedule. When daylight saving time shifts and you need every workflow running at 8 AM UTC, you update the manifest, commit it, and let the reconciler handle the rest.
The reconciler does the heavy lifting: it reads the manifest, queries the live Domo API for current schedule state, and computes a diff. Only when drift exists does it take action. This eliminates the noisy-log problem at the source. No drift, no writes, no activity log pollution — and a clean signal when something unexpected actually does change.
The GitHub Actions integration makes this production-ready. A scheduled workflow runs the reconciler on a set cadence. Critically, when drift is detected it doesn't auto-apply — it opens a Pull Request describing exactly what will change. That PR is your approval gate: a human reviews the proposed diff before anything lands in Domo. This is what separates a cron job from a real IaC workflow.
The apply step is explicitly idempotent — running it twice produces the same result as running it once. That matters when you're operating across many instances and someone inevitably triggers the workflow twice.
The pattern Jay demonstrates here extends well beyond schedule management. Any Domo configuration you want version-controlled — allow lists, user assignments, connector settings — fits the same shape: manifest → reconcile → PR → apply.
