Observability & APM Integration

During my load test, I observed the following in JMeter:
- Spike in p99 response time at [time] during [VU count]
- Error rate increased from 0% to [N]% at [time]

I have [Datadog / Dynatrace / New Relic / Grafana + Prometheus] as my APM tool.

Guide me on:
- Which APM dashboards and metrics to check
- How to find the slow traces correlating to the JMeter spike time
- Database query analysis: slow queries, lock contention
- JVM analysis: heap usage, GC activity, thread states
- Infrastructure correlation: CPU, memory, network at the same timestamp
- How to set up APM alerts triggered during load tests

Set up a complete real-time monitoring stack for JMeter:

Components:
- JMeter Backend Listener → InfluxDB → Grafana

Provide:
1. Docker Compose file for InfluxDB 2.x and Grafana
2. InfluxDB bucket and token setup commands
3. JMeter Backend Listener XML configuration
4. Grafana data source configuration
5. Key panels to add: throughput, response times (p50/p90/p95/p99), active VUs, error rate, requests/sec per transaction
6. How to annotate the Grafana dashboard with test start/end events

My Spring Boot application exposes Prometheus metrics via /actuator/prometheus.

During load testing, I want to correlate:
- JVM heap usage (jvm_memory_used_bytes)
- GC pause duration (jvm_gc_pause_seconds)
- Hikari connection pool active connections (hikaricp_connections_active)
- HTTP request duration from app side (http_server_requests_seconds)

Generate:
- Prometheus scrape config for the application
- Key PromQL queries to run during tests
- Grafana panel JSON for each metric
- Alert thresholds to set for each metric during a load test