Component Monitoring

Component monitoring collects metrics, traces, and logs from databases, caches, message queues, proxies, and containers using the OpenTelemetry Collector. Each guide configures a dedicated receiver or Prometheus scrape target and ships telemetry to base14 Scout.

Components by Category

Databases

Component	Guide	Key Metrics
PostgreSQL Basic	PostgreSQL Basic	Connections, query performance, locks, WAL
PostgreSQL Advanced	PostgreSQL Advanced	Query stats, table/index sizes, replication
MySQL	MySQL	Connections, queries, InnoDB, replication
MongoDB	MongoDB	Operations, connections, document metrics, cursors
Cassandra	Cassandra	Client requests, compaction, storage, caches
CouchDB	CouchDB	Request rates, document operations, view stats
Elasticsearch	Elasticsearch	Cluster health, node stats, JVM, index operations
ClickHouse	ClickHouse	Queries, inserts, memory tracking, merge operations
Couchbase	Couchbase	Cluster management, KV connections, CPU, memory
MariaDB	MariaDB	Connections, queries, InnoDB, replication
Materialize	Materialize	Dataflow freshness, source lag, replica memory, peeks
Trino	Trino	Query lifecycle, cluster memory pools, execution slots, query traces

Time-Series Databases

Component	Guide	Key Metrics
InfluxDB	InfluxDB	Write throughput, query duration, storage, cardinality

Key-Value & Distributed Storage

Component	Guide	Key Metrics
Aerospike	Aerospike	Connections, transactions, memory, namespace stats
etcd	etcd	Raft proposals, disk latency, MVCC, gRPC

Vector Databases

Component	Guide	Key Metrics
Qdrant	Qdrant	Request rate and latency, collection points, update queue, memory and mmap ceilings
Milvus	Milvus	Request rate and error rate, search and insert latency, ingestion lag, entity counts, segment growth
Weaviate	Weaviate	REST, GraphQL and gRPC request rate and errors, query and write latency, vector index size, async index queue depth

Search

Component	Guide	Key Metrics
Solr	Solr	JVM heap, GC, request rates, cores, threads
OpenSearch	OpenSearch	Cluster health, search latency, JVM, storage I/O

Caching

Component	Guide	Key Metrics
Redis	Redis	Memory, keyspace, commands, clients, replication
Memcached	Memcached	Hit ratio, memory, connections, evictions
Varnish	Varnish	Cache hit/miss, backend health, connections

Message Queues

Component	Guide	Key Metrics
RabbitMQ	RabbitMQ	Queue depth, message rates, node memory, I/O
Kafka	Kafka	Consumer lag, partition offsets, broker count
NATS	NATS	Connections, subscriptions, message rates, JetStream
Pulsar	Pulsar	Broker throughput, backlog, managed ledger, storage
ActiveMQ	ActiveMQ	Queue depth, enqueue/dequeue, producers, consumers
Redpanda	Redpanda	Produce/fetch throughput, partition health, Raft leadership

Service Discovery & Coordination

Component	Guide	Key Metrics
Consul	Consul	Raft consensus, service catalog, RPC, gossip
ZooKeeper	ZooKeeper	Connections, latency, znodes, watches, packets

Secrets Management

Component	Guide	Key Metrics
Vault	Vault	Seal operations, token lifecycle, barrier, leases

AI Model Serving

Component	Guide	Key Metrics
vLLM	vLLM	KV-cache usage, request queueing, token throughput, latency phases
llama.cpp	llama.cpp	Slot occupancy and queueing, token throughput, prompt-cache reuse, batching efficiency

AI Gateways

Component	Guide	Key Metrics
LiteLLM Gateway	LiteLLM Gateway	Deployment health and cooldowns, request failures, latency split, token spend
Bifrost	Bifrost	LLM request outcomes, provider latency, token usage, streaming performance

AI Agent Runtimes

Component	Guide	Key Metrics
OpenClaw	OpenClaw	Run and model call latency, token usage, lane queueing, context size; traces and logs
Claude Code	Claude Code	Cost and token usage per model, sessions, edit decisions, lines of code; traces and logs
Codex CLI	Codex CLI	Turn duration, tokens per turn, model requests, tool calls; traces and logs

Distributed Compute

Component	Guide	Key Metrics
Ray	Ray	Task and actor state, cluster resources, object store and spilling, scheduler placement

Orchestration

Component	Guide	Key Metrics
Airflow	Apache Airflow	Scheduler health, task outcomes, pool and executor saturation, DAG run traces
dbt	dbt	Run count, failed models and tests, model build time, run traces
Temporal	Temporal	Workflow latency, task queues, persistence, shards
Hatchet	Hatchet	Task inflow and outcome, queue backlog, worker slot capacity, per-workflow duration
Restate	Restate	Invocation rate and outcome, partition health, invoker backlog, workflow traces
Nomad	Nomad	Raft consensus, broker, RPC, job status, autopilot

Continuous Delivery

Component	Guide	Key Metrics
ArgoCD	ArgoCD	App sync status, health, reconciliation, Git ops
Jenkins	Jenkins	Build results, executor usage, queue depth

Web Servers & Proxies

Component	Guide	Key Metrics
NGINX	NGINX	Connections, request rate, worker states
Apache HTTP Server	Apache HTTP Server	Workers, scoreboard, request rate, bytes transferred
HAProxy	HAProxy	Sessions, request rate, backend health, queue depth
Traefik	Traefik	Entrypoint requests, TLS, router stats, open conns
Envoy	Envoy	Downstream connections, listeners, cluster manager
Caddy	Caddy	Request rates, response codes, TLS handshakes

Object Storage

Component	Guide	Key Metrics
MinIO	MinIO	Cluster capacity, drive health, S3 requests, ILM

Database Proxies

Component	Guide	Key Metrics
PgBouncer	PgBouncer	Connection pools, query throughput, client wait time

Containers

Component	Guide	Key Metrics
Docker Engine	Docker Engine	CPU, memory, block I/O, network per container
containerd	containerd	CRI errors, image pull failures, container lifecycle latency, CPU and memory per container
K3s	K3s	API server errors, container start failures, certificate expiry, node headroom

Java Application Servers

Component	Guide	Key Metrics
Tomcat	Tomcat	Request rates, thread pools, sessions, network I/O
Jetty	Jetty	Threads, I/O selects, sessions, request queue
WildFly	WildFly	Undertow requests, datasource pools, transactions

Network & IoT Devices

Component	Guide	Key Metrics
SNMP	SNMP	Interface I/O, CPU/memory, UPS battery, device status

For MQTT, Sparkplug B, OPC-UA, and edge Collector store-and-forward patterns, see IoT & Edge Instrumentation.

How Component Monitoring Works

Each component exposes metrics through one of four methods:

1. Dedicated OTel receiver - the Collector connects directly to the component's stats API (PostgreSQL, MySQL, MariaDB, MongoDB, Redis, RabbitMQ, Elasticsearch, CouchDB, Memcached, Apache HTTP Server, HAProxy, ZooKeeper, Kafka, Aerospike, Docker Engine)
2. Prometheus scrape - the component or a sidecar exporter exposes a /metrics endpoint that the Collector scrapes (Cassandra via JMX exporter, Consul, Vault, etcd, Solr, Temporal, NGINX, ClickHouse, NATS via prometheus-nats-exporter, Traefik, Envoy, MinIO, OpenSearch via prometheus-exporter plugin, PgBouncer via pgbouncer-exporter, Nomad, Couchbase, Pulsar, ArgoCD, Jenkins via Prometheus Metrics plugin, InfluxDB, Caddy, Varnish via prometheus_varnish_exporter, Redpanda via its native /public_metrics endpoint, Materialize via its native /metrics/public endpoint, Trino via its native /metrics endpoint, containerd via its native /v1/metrics endpoint, K3s via the kubelet's /metrics endpoint)
3. JMX Scraper - a standalone process connects to the application's JMX port via RMI, converts MBeans to OpenTelemetry metrics, and exports OTLP to the Collector (Tomcat, ActiveMQ, Jetty, WildFly). See JMX Metrics Collection Guide.
4. Native OTLP push - the component builds an OpenTelemetry SDK into its own processes and pushes to the Collector, so there is no endpoint to scrape (Apache Airflow, dbt).

Java applications using JMX have two collection approaches: the OTel JMX Scraper (remote, OTLP-native) and the Prometheus JMX Exporter (in-process agent). See JMX Metrics Collection Guide for a detailed comparison.

NGINX also supports distributed traces via nginx-module-otel and log collection via the filelog receiver.

Next Steps

1. Choose your component from the tables above
2. Follow the guide to configure the OTel Collector receiver
3. Create dashboards in Scout - see Create Your First Dashboard