Skip to content
Flow Docs
Flow home Back to the main Flow site

Tutorial: scheduled fleet scan

This tutorial turns a one-off scan flow into a scheduled, fleet-wide operation. It follows the pattern from the fleet-scans case study. You run the same vetted scan across the whole estate on a cadence, and you review exceptions instead of green runs.

What you’ll build

Section titled “What you’ll build”

You build a scan template that is cloned once per target, and each clone carries a monthly schedule. Failures route to a diagnostic branch automatically. Passing runs need no attention.

  1. Start from the template. Open the Template Hub and download the scan template (or save your own validated scan flow as a template). The template embeds the full graph. It fetches the toolkit, runs the scan, and downloads the results, with a .fail branch that downloads the diagnostic output.

  2. Parameterize per target. Clone the template once per LPAR or target system and set the connection and high-level qualifier on each clone. You can also use a set-variable utility node at the head of the flow so the parameters are inputs rather than edits.

  3. Verify the failure branch. The diagnostic download must be reached only through the scan node’s .fail edge. It should have one inbound edge and no unconditional path, so it runs exactly when there is something to diagnose.

  4. Add the schedule. With the flow saved, open the Schedule toggle in the canvas toolbar. Pick the monthly preset (or a cron expression), set the IANA timezone, and choose the run-once catch-up policy so a missed window gets a single make-up run. Preview the next firings before saving.

  5. Let the scheduler own it. The background scheduler fires each due clone. It runs in Flow Studio, in Flow Server, or as flow schedule daemon on a utility host. Scheduled runs are tagged in History and tracked per schedule, with run count, last status, and next run. They also appear in the central Schedules view.

  6. Review exceptions only. A failing run keeps its diagnostic output and shows up in History’s scheduled filter. A passing run needs nothing from you. When the scan procedure changes, update the template once, and every clone picks it up on its next scheduled run.

Why this can’t drift

Section titled “Why this can’t drift”

Because the toolkit or artifact is fetched fresh and version-pinned every run, no target can quietly fall behind. The classic failure structurally can’t happen. That failure is a back-level local copy that produces silently-wrong results, which someone discovers weeks later, one ticket at a time.

The CLI variant

Section titled “The CLI variant”
Terminal window
flow schedule add fleet-scan-lpar1 --frequency monthly --timezone America/New_York
flow schedule add fleet-scan-lpar2 --frequency monthly --timezone America/New_York
flow schedule list
flow schedule daemon     # run the scheduler loop on this host
Section titled “Related”