Enable NGINX proxy cache only for responses with a safe identity key
NGINX proxy cache: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Set NGINX proxy timeouts from phase-specific latency evidence
NGINX proxy timeouts: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Add NGINX rate limiting without punishing every user behind one proxy
NGINX rate limiting: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Configure NGINX real client IP without trusting spoofed headers
NGINX real client IP: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Log NGINX request and upstream timing without high-cardinality chaos
NGINX request time logging: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Test and reload NGINX without replacing healthy workers blindly
Nginx test config reload: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Proxy WebSockets through NGINX with explicit upgrade handling
NGINX WebSocket proxy: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Use openssl s_client to test SNI, chain, and protocol explicitly
Openssl s_client TLS test: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Run pg_dump in production with consistency, load, and restore proof
Pg_dump production backup: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Read pg_stat_activity without confusing idle sessions with blockers
Pg_stat_activity: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Use pg_stat_statements to rank cumulative PostgreSQL query cost
Pg_stat_statements: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Run PostgreSQL ANALYZE when estimates are wrong, not as a ritual
PostgreSQL ANALYZE statistics: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
pg_cancel_backend vs pg_terminate_backend: choose the smaller interruption
Pg_cancel_backend vs pg_terminate_backend: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Choose PostgreSQL sslmode without disabling identity verification
PostgreSQL sslmode: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Run EXPLAIN ANALYZE in production without surprising writes or load
EXPLAIN ANALYZE production: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Build a PostgreSQL lock tree from blockers to waiting sessions
PostgreSQL lock tree query: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Monitor PostgreSQL replication slots before WAL fills the disk
PostgreSQL replication slots monitoring: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Choose a Prometheus alert for duration from failure tolerance
Prometheus alert for duration: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Use Prometheus blackbox exporter to test the user-visible path
Prometheus blackbox exporter: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Set a Prometheus cardinality budget before adding labels
Prometheus cardinality budget: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Calculate latency with histogram_quantile without averaging percentiles
Prometheus histogram_quantile: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Use Prometheus recording rules for stable, expensive shared queries
Prometheus recording rules: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Test Prometheus alerting rules with promtool before loading them
Promtool test rules: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.
Handle Prometheus missing metrics without converting absence into zero
Prometheus missing metrics staleness: inspect the effective state, interpret expected and abnormal results, make one bounded change, verify, and roll back safely.