Keep the current build local-first
A reproducible local lab makes service boundaries and failure behavior inspectable without claiming an OCI environment that was not deployed.
ENGINEERING PROJECT / 05
A local-first infrastructure engineering lab focused on service isolation, observability, controlled failure testing and secure site-to-site routing. The current implementation is a reproducible local lab, not a deployed hybrid-cloud environment.
01 / CONTEXT
THE WHYNexusCloud began as a hybrid-cloud design exercise connecting an on-premises virtual data center to Oracle Cloud Infrastructure. Practical account and local virtualization constraints made an end-to-end cloud deployment unreliable, so the scope was narrowed to a working local environment that can be started, inspected and tested repeatedly.
02 / SYSTEM DESIGN
THE STRUCTUREA high-level view of the documented workflow and component boundaries.
Two Nginx sites run as separate containers on independent Docker bridge networks.
Prometheus scrapes service and host metrics; Blackbox Exporter probes both sites for HTTP availability.
Grafana is provisioned with a Prometheus data source and dashboard. Prometheus alert rules cover site availability and missing targets.
Node Exporter exposes host metrics to the monitoring stack.
Linux network namespaces model two site routers and two clients connected by WireGuard. This exercise is separate from the Docker application networks.
03 / ENGINEERING LOG
THE BUILD SEQUENCEA transparent sequence of engineering milestones. Status labels describe documented work—not a live completion percentage.
Retained the infrastructure and networking learning goals while documenting that OCI provisioning and the broader enterprise design remain future work.
Used Docker Compose to run two Nginx sites on separate bridge networks with local-only published ports and container health checks.
Connected Prometheus, Blackbox Exporter, Node Exporter and Grafana with provisioned dashboards and alert rules.
Added a controlled failure-injection workflow that stops a selected site, checks that its alert fires, restores the container and verifies recovery.
Created isolated Linux router and client namespaces with a WireGuard tunnel and tests for bidirectional client traffic and prefix restrictions.
04 / DESIGN JUDGMENT
TRADE-OFFS & REASONSThe engineering choices that shape the system—and the reason each boundary exists.
A reproducible local lab makes service boundaries and failure behavior inspectable without claiming an OCI environment that was not deployed.
The WireGuard namespaces demonstrate routed site-to-site connectivity independently; the Docker application containers do not traverse the tunnel.
Stopping and restoring a test site checks both alert firing and recovery instead of relying only on a static dashboard screenshot.
The Grafana password is supplied through a local .env file, while generated WireGuard keys and populated peer configurations are stored outside the repository.
06 / IMPLEMENTATION
THE HOWThe work is broken into practical building blocks, from initial design through to validation.
Runs two Nginx application sites on separate Docker bridge networks.
Uses Prometheus and Blackbox Exporter to monitor HTTP availability and Node Exporter for host metrics.
Provides Grafana dashboards and provisioned Prometheus alert rules.
Tests controlled site outages, alert firing, service recovery and alert clearing.
Models two site routers and two clients in Linux network namespaces connected by WireGuard.
Uses PowerShell and Bash scripts plus GitHub Actions for repeatable validation.
07 / VERIFICATION
THE CHECKSSpecific checks documented for this project. Results are scoped to the recorded lab or test run.
Two Nginx sites are attached to separate Docker bridge networks and publish only to localhost.
The README documents four expected Prometheus scrape targets as the monitoring baseline.
The documented test stops a site, waits for its alert, restores it and checks that the probe and alert recover.
The WireGuard lab validates client-to-client traffic, tunnel handshake and allowed-prefix restrictions.
The repository documents Compose checks, shell syntax checks, service smoke tests, Prometheus checks and Linux network/port validation.
08 / RESULTS
THE VALUEA reproducible local infrastructure stack with service separation and centralized observability.
A controlled way to demonstrate monitoring, alert firing and recovery.
A separate WireGuard routing simulation with explicit network namespaces and prefix restrictions.
Operational scripts, documentation and CI checks that make the lab easier to repeat.
10 / ENGINEERING NOTES
TRANSPARENCYClear boundaries matter. These notes describe what the current project does—and what it does not claim to do.