MacDaddy Suite Case Study | Edmund Dionisio

01 / Problem

CLI evidence was scattered across too many manual steps.

This started with the kind of ticket every network engineer knows: a customer reports a device problem but can only provide a fragment of useful information. Finding that endpoint across more than 100 switches meant repeating gateway searches, MAC-table checks, path tracing, interface diagnostics, and follow-up commands by hand.

MacDaddy turns that sequence into one coordinated job. Automating discovery reduced response time and gave me one place to preserve the evidence, policy decision, change preview, verification, and rollback trail.

From my build notes

The MAC-tracing algorithm took the most trial and error. The real switching environment had paths and edge cases that did not fit a textbook diagram, so I kept instrumenting each hop, reading the logs, and tuning the approach until it matched how this network actually behaves.

DiscoverGateway-first MAC tracing and downstream path walking

DecideLive interface evidence and server-side safety verdicts

RecoverRollback snapshots, read-back checks, and history

02 / Architecture

A shared platform with a narrow device boundary.

The web interface coordinates jobs and presents evidence; device access remains behind typed validation, policy checks, bounded SSH resources, and a dedicated execution layer.

React UI

Search, evidence cards, previews, live job progress, reports, access controls.

FastAPI

Authenticated routes, CSRF protection, WebSockets, job and schedule coordination.

Policy + jobs

Target allowlists, safety verdicts, write leases, stale-evidence gates, audit events.

Jump host

Bounded two-hop SSH with retry/backoff and circuit-breaker behavior.

Network edge

Live read, guarded command execution, read-back verification, rollback capture.

03 / Workflow

Evidence drives every transition.

Normalize the query

Accept a full or partial MAC and validate the input before any device access.

Search gateways first

Reduce unnecessary access-switch sessions, then follow the learned downstream path.

Build a live interface record

Collect configuration, status, counters, authentication, neighbors, and learned MAC evidence.

Preview a guarded profile

Re-read live state, validate the proposed command scope, and show exactly what would change.

Apply, verify, and retain

Capture rollback, acquire a per-switch lease, execute, read back, save once per verified switch, and audit the result.

04 / Product evidence

The working interface keeps discovery, jobs, history, and recovery connected.

Sanitized MacDaddy operations dashboard — The dashboard makes inventory health, scan signal, remediation outcomes, and warnings visible in one place.

Sanitized MacDaddy multi-device scan workspace — The scan workflow starts from the limited identifiers that often arrive in a support ticket.

Sanitized MacDaddy live activity stream — Detailed activity logging exposes progress, failures, reports, and operator actions instead of hiding work behind a spinner.

Sanitized MacDaddy scheduled scan form — Scheduled scans reuse the same bounded scope, current permissions, and reporting path as manual work.

Sanitized MacDaddy device observation history — Retained observations turn one lookup into movement and location evidence over time.

Sanitized MacDaddy rollback preview — Rollback has its own preview, typed confirmation, verification, and audit path.

05 / Safety model

Write access is a policy decision, not a button state.

Preview and apply independently re-read device state. If required neighbor evidence fails, a target is stale, or the interface no longer satisfies policy, the operation stops closed.

Target allowlistsHosts and interfaces are validated at the API boundary.

Command scopeEvery profile command passes server-side validation.

Infrastructure protectionTrunks, port-channels, uplink signals, and infrastructure neighbors are refused.

Typed confirmationThe exact interface or safe-target count is required before apply.

Write leaseOverlapping writes to the same switch are prevented.

Rollback + verificationPre-change state is captured and the live result is read back.

Current permissionsScheduled work rechecks the creator's effective access at run time.

Secret redactionRuntime secrets and labeled values are filtered from application logs.

06 / Decisions

Hard constraints shaped the implementation.

01

Gateway-first discovery

Start where the MAC-routing evidence is most concentrated, then walk the path instead of opening sessions to every access switch.

02

One save per switch

Bulk work suppresses per-interface saves and writes memory once only after at least one target verifies.

03

Shared write choke point

Remediation and one-off interface tools reuse the same safety verdict, command validation, lease, rollback, verification, and audit path.

04

Staging fails closed

Isolated acceptance mode blocks live writes, schedules, and outbound email so test confidence does not depend on production reachability.

Verification evidence

The v80 source includes a structured local acceptance record and a broad automated test suite spanning parsers, policy, schedules, authorization, modules, and recovery behavior. This page describes the verified build without claiming a deployment date that has not been publicly confirmed.

07 / Evolution

Dionisio NES proved the workflow before the platform grew around it.

Dionisio NES established the useful core: inventory-driven lookup, path discovery, diagnostics, and guarded actions. MacDaddy retained those practical concepts, then rebuilt the experience around shared policy, job coordination, reporting, authorization, scheduling, and durable evidence.

Sanitized Dionisio NES terminal menu showing network lookup choices — The CLI-first foundation exposed focused lookup and diagnostic paths.

Sanitized Dionisio NES interface action workflow — Guarded operator actions became the basis for the suite's policy-driven write workflow.

08 / Lessons

Concurrency, stale state, and recovery matter more than the happy path.

The most important changes came from treating device access as a shared, failure-prone resource: bound the jump-host load, distinguish target failures from transport outages, invalidate old evidence, recover durable jobs, and keep every write on one guarded path.

The application remains environment-specific and requires authorized inventory and device access.
Legacy device compatibility is isolated rather than weakening the modern connection path globally.
Public screenshots use sanitized data; no public demo can reach live equipment.
Current-build acceptance and production deployment are reported separately.