Make recovery a deliberate operation
Use blue-green table swap patterns when recovering from full-text-search corruption.
Back to projectsProduction operations
Data Platform / Observability
Database Operations and Monitoring
Operates production MariaDB at scale and builds custom monitoring tooling around it, including a Go-based agent that exports metrics to Prometheus and InfluxDB.
Metrics flow
From database behavior to operational metrics
The custom monitoring path exports MariaDB metrics through a Go agent to Prometheus and InfluxDB. Recovery and query optimization are separate parts of the project scope.
Observe the database
Production behavior, locks and query performance.
- MariaDB
Collect metrics
Purpose-built database monitoring agent.
- Go
Export telemetry
Metrics output for operational monitoring.
- Prometheus
- InfluxDB
The challenge
Production database platforms need fast incident diagnosis, reliable recovery patterns, and metrics that expose operational risk before it becomes downtime.
The approach
Diagnosed lock convoys, recovered from full-text-search corruption with blue-green table swap patterns, optimized queries, planned upgrade work, and built a Go monitoring agent with Prometheus and InfluxDB output.
My role
Worked across production database troubleshooting, recovery planning, query optimization, and custom monitoring implementation.
Engineering decisions
Expose database-specific behavior
Build a Go agent with Prometheus and InfluxDB outputs to support production monitoring.
Scope & delivery
- MariaDB lock convoy diagnosis and production recovery work.
- Full-text-search corruption recovery through blue-green table swap patterns.
- Query optimization and major version upgrade planning.
- Go-based monitoring agent with Prometheus and InfluxDB outputs.