Running across a fleet¶

pydebuprep runs on one host at a time; to cover many, run it from your configuration management. Use JSON output, collect the files, and act on the exit code.

With Ansible¶

- name: Release upgrade with pydebuprep checks
  hosts: upgrade_targets
  become: true
  tasks:
    - name: Take a pre-upgrade snapshot
      ansible.builtin.command: pydebuprep snapshot
      changed_when: false

    # ... your upgrade and reboot tasks ...

    - name: Create report directory
      ansible.builtin.file:
        path: /var/tmp/pydebuprep
        state: directory
        mode: "0750"

    - name: Run pydebuprep
      ansible.builtin.command: >-
        pydebuprep report --format json --output-dir /var/tmp/pydebuprep
      register: pydebuprep
      changed_when: false
      failed_when: pydebuprep.rc == 3

    - name: Find the newest report
      ansible.builtin.find:
        paths: /var/tmp/pydebuprep
        patterns: "pydebuprep-*.json"
      register: reports

    - name: Fetch the report
      ansible.builtin.fetch:
        src: "{{ (reports.files | sort(attribute='mtime') | last).path }}"
        dest: "reports/{{ inventory_hostname }}.json"
        flat: true

    - name: Flag hosts with critical findings
      ansible.builtin.debug:
        msg: "{{ inventory_hostname }} has critical findings, see reports/{{ inventory_hostname }}.json"
      when: pydebuprep.rc == 2

Notes:

changed_when: false — pydebuprep never changes the host, so the task never reports a change.
failed_when: pydebuprep.rc == 3 — exit codes 1 and 2 mean findings, not a failure of the run. Only 3 means pydebuprep itself failed.
Distribute a common /etc/pydebuprep.conf (for example with ansible.builtin.copy) to silence findings that are expected everywhere, and check it with pydebuprep configtest.

Turning collected reports into something to read¶

The fetched JSON files can be converted on the control node, without touching the hosts again:

for f in reports/*.json; do
    pydebuprep convert "$f" --format html --output-dir reports/html
done

Add --todo to get each host's checklist, ready to paste into a ticket:

pydebuprep convert reports/web1.json --format jira --todo -o -

For dashboards, the report format gives each host's summary counts and exit code directly:

jq -r '[.host.hostname, .summary.critical, .summary.warning] | @tsv' reports/*.json