High-Availability Multi-Tenant Infrastructure & Memory Optimization

Co-locating multiple Django WSGI processes, PostgreSQL instances, and an Nginx reverse proxy on a 1GB RAM limit instantly triggered the Linux Out-Of-Memory (OOM) killer. The OS kernel was terminating active Gunicorn processes to free up RAM, leading to recurrent system crashes and application downtime.

1. Memory Overcommit Tuning: Allocated and configured an optimized Swap Space subsystem to absorb memory spikes during DB transactions.
2. WSGI Worker Restructuring: Scaled down from 3 workers to 1 master worker per app, significantly reducing context-switching latency and memory footprint.
3. Dedicated Reverse Proxying: Configured Nginx to handle SSL handshakes and route incoming requests to isolated socket points.
4. Independent Deployment Automation: Programmed distinct GitHub Actions runners to safely execute deployments on target branches.

To scale application hosting efficiently, this project involved designing an isolated multi-tenant architecture on a minimal resource budget. The objective was hosting three completely distinct production portfolios (separate codebases, isolated databases, and custom domains) on a single 1GB RAM Ubuntu VPS. Key Technical Highlights: Virtual Host Routing: Custom Nginx configuration as a high-performance reverse proxy, handling domain-to-socket translation for multiple WSGI entry points. Continuous Delivery: Independent branch-targeted GitHub Actions workflows to automate code pushes, enabling independent deployments with zero service disruption. Low-Footprint Process Architecture: Customized systemd service daemons optimized to run single Gunicorn worker threads, preventing RAM exhaustion.

1. Achieved 100% service uptime with zero OOM crashes under parallel traffic loads.
2. Realized 3x hosting cost savings by successfully hosting 3 distinct websites on a single $5/month instance.
3. Established total database and code isolation, allowing independent content management for each tenant.