Set up containers

I need three containerized services for the payments stack: an API on port 8080, a Redis cache, and a PostgreSQL 13 instance, each with resource limits (API 1 CPU/1Gi, Redis 0.5 CPU/512Mi, Postgres 2 CPU/4Gi). Deploy them into the same user namespace, mount persistent storage for Postgres, set restart policy on-failure, and give me the container IDs and the compose or manifest file so I can version it.

Before I hand these containers to QA, make the deployment reviewer-friendly: put the API's exposed port and expected health-check URL at the top of the manifest, surface the do-not-use-in-prod flags, make the volume mapping for Postgres easy to find, and flag any capabilities or host networking that would make a security reviewer hesitate.

I launched the payments containers with the requested limits and mounts; they pass health checks at idle but the API gives intermittent 500s when the load test ramps. I suspect resource contention or an unoptimized DB connection pool but can't tell which. What diagnostic steps will quickly show whether it's CPU throttling, memory OOM, or database saturation, and what immediate mitigation should I apply to keep QA running?

We keep losing time because containers that work locally fail in our shared environment—wrong resource limits, inconsistent env vars, and undocumented host mounts. I want to change one habit across the team to stop firefighting these issues. Should we require a single canonical manifest in the repo with enforced CI validation, a local emulation standard, or mandatory performance smoke tests before merge, and how do we get the team to comply?