setup-prometheus-monitoring
Acerca de
Esta habilidad configura un despliegue de Prometheus listo para producción para la recolección centralizada de métricas de series temporales. Establece configuraciones de scrape, descubrimiento de servicios, reglas de grabación y federación para entornos multi-clúster. Úsela para establecer una base de monitoreo para microservicios, infraestructura y seguimiento de SLOs al migrar a una plataforma de observabilidad moderna.
Instalación rápida
Claude Code
Recomendadonpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/setup-prometheus-monitoringCopia y pega este comando en Claude Code para instalar esta habilidad
Documentación
Setup Prometheus Monitoring
Configure a production-ready Prometheus deployment with scrape targets, recording rules, and federation.
When to Use
- Setting up centralized metrics collection for microservices or distributed systems
- Implementing time-series monitoring for application and infrastructure metrics
- Establishing a foundation for SLO/SLI tracking and alerting
- Consolidating metrics from multiple Prometheus instances via federation
- Migrating from legacy monitoring solutions to a modern observability stack
Inputs
- Required: List of scrape targets (services, exporters, endpoints)
- Required: Retention period and storage requirements
- Optional: Existing service discovery mechanism (Kubernetes, Consul, EC2)
- Optional: Recording rules for pre-aggregated metrics
- Optional: Federation hierarchy for multi-cluster setups
Procedure
Step 1: Install and Configure Prometheus
Create the base Prometheus configuration with global settings and scrape intervals.
# Create Prometheus directory structure
mkdir -p /etc/prometheus/{rules,file_sd}
mkdir -p /var/lib/prometheus
# Download Prometheus (adjust version as needed)
cd /tmp
wget https://github.com/prometheus/prometheus/releases/download/v2.48.0/prometheus-2.48.0.linux-amd64.tar.gz
tar xvf prometheus-2.48.0.linux-amd64.tar.gz
sudo cp prometheus-2.48.0.linux-amd64/{prometheus,promtool} /usr/local/bin/
Create /etc/prometheus/prometheus.yml:
global:
scrape_interval: 15s
scrape_timeout: 10s
evaluation_interval: 15s
external_labels:
cluster: 'production'
region: 'us-east-1'
# Alertmanager configuration
alerting:
alertmanagers:
- static_configs:
- targets:
- localhost:9093
# Load recording and alerting rules
rule_files:
- "rules/*.yml"
# Scrape configurations
scrape_configs:
# Prometheus self-monitoring
- job_name: 'prometheus'
static_configs:
- targets: ['localhost:9090']
labels:
env: 'production'
# Node exporter for host metrics
- job_name: 'node'
static_configs:
- targets:
- 'node1:9100'
- 'node2:9100'
labels:
env: 'production'
# Application metrics with file-based service discovery
- job_name: 'app-services'
file_sd_configs:
- files:
- '/etc/prometheus/file_sd/services.json'
refresh_interval: 30s
relabel_configs:
- source_labels: [__address__]
target_label: instance
- source_labels: [env]
target_label: environment
Got: Prometheus starts successfully, web UI accessible at http://localhost:9090, targets listed under Status > Targets.
If fail:
- Check syntax with
promtool check config /etc/prometheus/prometheus.yml - Verify file permissions:
sudo chown -R prometheus:prometheus /etc/prometheus /var/lib/prometheus - Check logs:
journalctl -u prometheus -f
Step 2: Configure Service Discovery
Set up dynamic target discovery to avoid manual target management.
For Kubernetes environments, add to scrape_configs:
- job_name: 'kubernetes-pods'
kubernetes_sd_configs:
- role: pod
relabel_configs:
# Only scrape pods with prometheus.io/scrape annotation
- source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_scrape]
action: keep
regex: true
# Use custom port if specified
- source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_port]
action: replace
target_label: __address__
regex: ([^:]+)(?::\d+)?;(\d+)
replacement: $1:$2
# Add namespace as label
- source_labels: [__meta_kubernetes_namespace]
target_label: kubernetes_namespace
# Add pod name as label
- source_labels: [__meta_kubernetes_pod_name]
target_label: kubernetes_pod_name
For file-based service discovery, create /etc/prometheus/file_sd/services.json:
[
{
"targets": ["web-app-1:8080", "web-app-2:8080"],
"labels": {
"job": "web-app",
"env": "production",
"team": "platform"
}
},
{
"targets": ["api-service-1:9090", "api-service-2:9090"],
"labels": {
"job": "api-service",
"env": "production",
"team": "backend"
}
}
]
For Consul service discovery:
- job_name: 'consul-services'
consul_sd_configs:
- server: 'consul.example.com:8500'
services: [] # Empty list means discover all services
relabel_configs:
- source_labels: [__meta_consul_service]
target_label: job
- source_labels: [__meta_consul_tags]
regex: '.*,monitoring,.*'
action: keep
Got: Dynamic targets appear in Prometheus UI, automatically updated when services scale or change.
If fail:
- Kubernetes: Verify RBAC permissions with
kubectl auth can-i list pods --as=system:serviceaccount:monitoring:prometheus - File SD: Validate JSON with
python -m json.tool /etc/prometheus/file_sd/services.json - Consul: Test connectivity with
curl http://consul.example.com:8500/v1/catalog/services
Step 3: Create Recording Rules
Pre-aggregate expensive queries for dashboard performance and alerting efficiency.
Create /etc/prometheus/rules/recording_rules.yml:
groups:
- name: api_aggregations
interval: 30s
rules:
# Calculate request rate per endpoint (5m window)
- record: job:http_requests:rate5m
expr: |
sum by (job, endpoint, method) (
rate(http_requests_total[5m])
)
# Calculate error rate percentage
- record: job:http_errors:rate5m
expr: |
sum by (job) (
rate(http_requests_total{status=~"5.."}[5m])
) / sum by (job) (
rate(http_requests_total[5m])
) * 100
# P95 latency by endpoint
- record: job:http_request_duration_seconds:p95
expr: |
histogram_quantile(0.95,
sum by (job, endpoint, le) (
rate(http_request_duration_seconds_bucket[5m])
)
)
- name: resource_aggregations
interval: 1m
rules:
# CPU usage by instance
- record: instance:cpu_usage:ratio
expr: |
1 - avg by (instance) (
rate(node_cpu_seconds_total{mode="idle"}[5m])
)
# Memory usage percentage
- record: instance:memory_usage:ratio
expr: |
1 - (
node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes
)
# Disk usage by mount point
- record: instance:disk_usage:ratio
expr: |
1 - (
node_filesystem_avail_bytes{fstype!~"tmpfs|fuse.*"}
/ node_filesystem_size_bytes{fstype!~"tmpfs|fuse.*"}
)
Validate and reload:
# Validate rules syntax
promtool check rules /etc/prometheus/rules/recording_rules.yml
# Reload Prometheus configuration (without restart)
curl -X POST http://localhost:9090/-/reload
# Or send SIGHUP signal
sudo killall -HUP prometheus
Got: Recording rules evaluate successfully, new metrics visible in Prometheus with job: prefix, query performance improved for dashboards.
If fail:
- Check rule syntax with
promtool check rules - Verify evaluation interval matches data availability
- Check for missing source metrics:
curl http://localhost:9090/api/v1/targets - Review logs for evaluation errors:
journalctl -u prometheus | grep -i error
Step 4: Configure Storage and Retention
Optimize storage for retention requirements and query performance.
Edit /etc/systemd/system/prometheus.service:
[Unit]
Description=Prometheus Monitoring System
Documentation=https://prometheus.io/docs/introduction/overview/
After=network-online.target
[Service]
Type=simple
User=prometheus
Group=prometheus
ExecStart=/usr/local/bin/prometheus \
--config.file=/etc/prometheus/prometheus.yml \
--storage.tsdb.path=/var/lib/prometheus \
--storage.tsdb.retention.time=30d \
--storage.tsdb.retention.size=50GB \
--web.console.templates=/etc/prometheus/consoles \
--web.console.libraries=/etc/prometheus/console_libraries \
--web.listen-address=:9090 \
--web.enable-lifecycle \
--web.enable-admin-api
Restart=always
RestartSec=10s
[Install]
WantedBy=multi-user.target
Key storage flags:
--storage.tsdb.retention.time=30d: Keep 30 days of data--storage.tsdb.retention.size=50GB: Limit storage to 50GB (whichever limit hits first)--storage.tsdb.wal-compression: Enable WAL compression (reduces disk I/O)--web.enable-lifecycle: Allow config reload via HTTP POST--web.enable-admin-api: Enable snapshot and delete APIs
Enable and start:
sudo systemctl daemon-reload
sudo systemctl enable prometheus
sudo systemctl start prometheus
sudo systemctl status prometheus
Got: Prometheus retains metrics according to policy, disk usage stays within limits, old data automatically pruned.
If fail:
- Monitor disk usage:
du -sh /var/lib/prometheus - Check TSDB stats:
curl http://localhost:9090/api/v1/status/tsdb - Verify retention settings:
curl http://localhost:9090/api/v1/status/runtimeinfo | jq .data.storageRetention - Force cleanup:
curl -X POST http://localhost:9090/api/v1/admin/tsdb/delete_series?match[]={__name__=~".+"}
Step 5: Set Up Federation (Multi-Cluster)
Configure hierarchical Prometheus for aggregating metrics across clusters.
On edge Prometheus instances (in each cluster), ensure external labels are set:
global:
external_labels:
cluster: 'production-east'
datacenter: 'us-east-1'
On central Prometheus instance, add federation scrape config:
scrape_configs:
- job_name: 'federate-production'
honor_labels: true
metrics_path: '/federate'
params:
'match[]':
# Aggregate only pre-computed recording rules
- '{__name__=~"job:.*"}'
# Include alert states
- '{__name__=~"ALERTS.*"}'
# Include critical infrastructure metrics
- 'up{job=~".*"}'
static_configs:
- targets:
- 'prometheus-east.example.com:9090'
- 'prometheus-west.example.com:9090'
labels:
env: 'production'
relabel_configs:
- source_labels: [__address__]
target_label: instance
- source_labels: [__address__]
regex: 'prometheus-(.*).example.com.*'
target_label: cluster
replacement: '$1'
Federation best practices:
- Use
honor_labels: trueto preserve original labels - Federate only recording rules and aggregates (not raw metrics)
- Set appropriate scrape intervals (longer than edge Prometheus evaluation)
- Use
match[]to filter metrics (avoid federating everything)
Got: Central Prometheus shows federated metrics from all clusters, queries can span multiple regions, minimal data duplication.
If fail:
- Verify federation endpoint accessibility:
curl http://prometheus-east.example.com:9090/federate?match[]={__name__=~"job:.*"} | head -20 - Check for label conflicts (central vs edge external labels)
- Monitor federation lag: compare timestamp differences
- Review match patterns:
curl http://localhost:9090/api/v1/label/__name__/values | jq .data | grep "job:"
Step 6: Implement High Availability (Optional)
Deploy redundant Prometheus instances with identical configurations for failover.
Use Thanos or Cortex for true HA, or simple load-balanced setup:
# prometheus-1.yml and prometheus-2.yml (identical configs)
global:
scrape_interval: 15s
external_labels:
prometheus: 'prometheus-1' # Different per instance
replica: 'A'
# Use --web.external-url flag for each instance
# prometheus-1: --web.external-url=http://prometheus-1.example.com:9090
# prometheus-2: --web.external-url=http://prometheus-2.example.com:9090
Configure Grafana to query both instances:
{
"name": "Prometheus-HA",
"type": "prometheus",
"url": "http://prometheus-lb.example.com",
"jsonData": {
"httpMethod": "POST",
"timeInterval": "15s"
}
}
Use HAProxy or nginx for load balancing:
upstream prometheus_backend {
server prometheus-1.example.com:9090 max_fails=3 fail_timeout=30s;
server prometheus-2.example.com:9090 max_fails=3 fail_timeout=30s;
}
server {
listen 9090;
location / {
proxy_pass http://prometheus_backend;
proxy_set_header Host $host;
}
}
Got: Query requests balanced across instances, automatic failover if one instance down, no data loss during single instance failure.
If fail:
- Verify both instances scraping same targets (slight time skew acceptable)
- Check for configuration drift between instances
- Monitor deduplication in queries (Grafana shows duplicate series)
- Review load balancer health checks
Validation
- Prometheus web UI accessible at expected endpoint
- All configured scrape targets showing as UP in Status > Targets
- Service discovery dynamically adding/removing targets as expected
- Recording rules evaluating successfully (no errors in logs)
- Metrics retention matching configured time/size limits
- Federation (if configured) pulling metrics from edge instances
- Queries returning expected metric cardinality (not excessive)
- Disk usage stable and within allocated storage budget
- Configuration reload working via HTTP endpoint or SIGHUP
- Prometheus self-monitoring metrics available (up, scrape duration, etc.)
Pitfalls
- High cardinality metrics: Avoid labels with unbounded values (user IDs, timestamps, UUIDs). Use recording rules to aggregate before storage.
- Scrape interval mismatch: Recording rules should evaluate at intervals equal to or greater than scrape intervals to avoid gaps.
- Federation overload: Federating all metrics creates massive data duplication. Only federate aggregated recording rules.
- Missing relabel configs: Without proper relabeling, service discovery can create confusing or duplicate labels.
- Retention too short: Set retention longer than your longest dashboard time window to avoid "no data" gaps.
- No resource limits: Prometheus can consume excessive memory with high cardinality. Set
--storage.tsdb.max-block-durationand monitor heap usage. - Disabled lifecycle endpoint: Without
--web.enable-lifecycle, config reloads require full restarts causing scrape gaps.
Related Skills
configure-alerting-rules- Define alerting rules based on Prometheus metrics and route to Alertmanagerbuild-grafana-dashboards- Visualize Prometheus metrics with Grafana dashboards and panelsdefine-slo-sli-sla- Establish SLO/SLI targets using Prometheus recording rules and error budget trackinginstrument-distributed-tracing- Complement metrics with distributed tracing for deeper observability
Repositorio GitHub
Habilidades relacionadas
llamaguard
OtroLlamaGuard es el modelo de Meta de 7-8B parámetros para moderar las entradas y salidas de LLM en seis categorías de seguridad como violencia y discurso de odio. Ofrece una precisión del 94-95% y puede implementarse usando vLLM, Hugging Face o Amazon SageMaker. Utiliza esta skill para integrar fácilmente filtrado de contenido y barreras de seguridad en tus aplicaciones de IA.
cost-optimization
OtroEsta Skill de Claude ayuda a los desarrolladores a optimizar los costes en la nube mediante el ajuste de tamaño de recursos, estrategias de etiquetado y análisis de gastos. Proporciona un marco para reducir los gastos en la nube e implementar una gobernanza de costes en AWS, Azure y GCP. Úsala cuando necesites analizar los costes de infraestructura, ajustar el tamaño de los recursos o cumplir con restricciones presupuestarias.
quantizing-models-bitsandbytes
OtroEsta habilidad cuantiza LLMs a precisión de 8 o 4 bits utilizando bitsandbytes, logrando una reducción de memoria del 50-75% con pérdida mínima de precisión. Es ideal para ejecutar modelos más grandes en memoria GPU limitada o para acelerar la inferencia, admitiendo formatos como INT8, NF4 y FP4. La habilidad se integra con HuggingFace Transformers y permite entrenamiento QLoRA y optimizadores de 8 bits.
dispatching-parallel-agents
OtroEsta Skill de Claude despliega múltiples agentes para investigar y solucionar 3 o más problemas independientes de forma concurrente. Está diseñada para escenarios que involucran fallos no relacionados que pueden resolverse sin estado compartido o dependencias. Su capacidad principal es la resolución paralela de problemas, asignando un agente por cada dominio problemático independiente para maximizar la eficiencia.
