The client keeps production control.
No credentials or production access required: your team runs read-only collection and shares the resulting evidence bundle.
No credentials or production access required: your team runs read-only collection and shares the resulting evidence bundle.
Define the production symptom, business impact and evidence boundary.
Client-run read-only evidence: workload, plans, locks, indexes and relevant settings.
Correlate signals and distinguish the symptom from likely root cause.
Rank findings by impact, confidence, risk and implementation effort.
Define rollback and before/after checks so the team knows whether the change worked.
pg_monitor role, once in a normal period and, if possible, once while the problem is happening. Add representative EXPLAIN (ANALYZE, BUFFERS) output from a safe environment, a short change timeline and, on RDS or Aurora, Database Insights and CloudWatch exports for the same window.default_transaction_read_only = on, a 30-second statement_timeout and a 1-second lock_timeout for its own session, and reads only statistics and catalog views.pg_stat_statements text is normalized, so literal parameter values are replaced with $1, $2, ….It reduces the security hurdle, keeps credentials inside your environment and makes the evidence bundle inspectable before it leaves your control. Read the collector script.
Because the decision may depend on workload frequency, writes, locks, connections, statistics, platform limits and the cost of changing the wrong thing.
Start with the problem and the evidence you already have. You will get a direct answer on whether a scoped investigation is the right next step.
Prefer to talk first? Book a 20-minute triage call.