aula-12: Victoria Metrics + Grafana via GitOps

Stack completo de observabilidade: - Victoria Metrics (vmsingle, vmagent, vmalert) - Grafana com dashboards built-in - Alertas customizados (PVC, pods, nodes, deployments) - pvc-autoresizer para expansão automática de volumes - Queries PromQL documentadas Instalação via ArgoCD seguindo padrão GitOps da aula-11.
2026-01-08 17:11:28 -03:00
parent e75b245c3b
commit 4b92838ac3
9 changed files with 1939 additions and 0 deletions
--- a/aula-12/queries/useful-queries.md
+++ b/aula-12/queries/useful-queries.md
@@ -0,0 +1,323 @@
+# Queries PromQL Úteis
+
+Queries prontas para uso no Grafana ou diretamente na API do Victoria Metrics.
+
+## Como usar
+
+### Via Grafana
+1. Acesse Grafana → Explore
+2. Selecione datasource "VictoriaMetrics"
+3. Cole a query no editor
+
+### Via API
+```bash
+# Port-forward
+kubectl port-forward -n monitoring svc/vmsingle-vm-victoria-metrics-k8s-stack 8429:8429
+
+# Query
+curl "http://localhost:8429/api/v1/query?query=up"
+```
+
+---
+
+## Storage / PVC
+
+### Uso de PVC em porcentagem
+```promql
+kubelet_volume_stats_used_bytes / kubelet_volume_stats_capacity_bytes * 100
+```
+
+### PVCs acima de 80%
+```promql
+(kubelet_volume_stats_used_bytes / kubelet_volume_stats_capacity_bytes) > 0.8
+```
+
+### Espaço disponível por PVC (bytes)
+```promql
+kubelet_volume_stats_available_bytes
+```
+
+### Espaço disponível por PVC (GB)
+```promql
+kubelet_volume_stats_available_bytes / 1024 / 1024 / 1024
+```
+
+### Inodes disponíveis
+```promql
+kubelet_volume_stats_inodes_free / kubelet_volume_stats_inodes * 100
+```
+
+### PVCs que vão encher em 24h (previsão)
+```promql
+predict_linear(kubelet_volume_stats_available_bytes[6h], 24 * 3600) < 0
+```
+
+---
+
+## CPU
+
+### CPU por pod (cores)
+```promql
+sum(rate(container_cpu_usage_seconds_total{container!=""}[5m])) by (pod, namespace)
+```
+
+### CPU por namespace (cores)
+```promql
+sum(rate(container_cpu_usage_seconds_total{container!=""}[5m])) by (namespace)
+```
+
+### CPU por node (%)
+```promql
+100 - (avg by(instance) (rate(node_cpu_seconds_total{mode="idle"}[5m])) * 100)
+```
+
+### Top 10 pods por CPU
+```promql
+topk(10, sum(rate(container_cpu_usage_seconds_total{container!=""}[5m])) by (pod, namespace))
+```
+
+### Uso de CPU vs Request
+```promql
+sum(rate(container_cpu_usage_seconds_total{container!=""}[5m])) by (pod, namespace)
+/
+sum(kube_pod_container_resource_requests{resource="cpu"}) by (pod, namespace)
+```
+
+---
+
+## Memória
+
+### Memória por pod (bytes)
+```promql
+sum(container_memory_working_set_bytes{container!=""}) by (pod, namespace)
+```
+
+### Memória por namespace (GB)
+```promql
+sum(container_memory_working_set_bytes{container!=""}) by (namespace) / 1024 / 1024 / 1024
+```
+
+### Memória disponível por node (%)
+```promql
+(node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes) * 100
+```
+
+### Top 10 pods por memória
+```promql
+topk(10, sum(container_memory_working_set_bytes{container!=""}) by (pod, namespace))
+```
+
+### Uso de memória vs Limit
+```promql
+sum(container_memory_working_set_bytes{container!=""}) by (pod, namespace)
+/
+sum(kube_pod_container_resource_limits{resource="memory"}) by (pod, namespace)
+```
+
+---
+
+## Pods e Containers
+
+### Pods restartando na última hora
+```promql
+sum(increase(kube_pod_container_status_restarts_total[1h])) by (pod, namespace) > 0
+```
+
+### Pods não Ready
+```promql
+kube_pod_status_ready{condition="false"}
+```
+
+### Pods em CrashLoopBackOff
+```promql
+kube_pod_container_status_waiting_reason{reason="CrashLoopBackOff"}
+```
+
+### Pods pendentes
+```promql
+kube_pod_status_phase{phase="Pending"}
+```
+
+### Containers OOMKilled
+```promql
+kube_pod_container_status_last_terminated_reason{reason="OOMKilled"}
+```
+
+### Total de pods por namespace
+```promql
+sum(kube_pod_info) by (namespace)
+```
+
+### Pods por node
+```promql
+sum(kube_pod_info) by (node)
+```
+
+---
+
+## Deployments
+
+### Deployments com réplicas indisponíveis
+```promql
+kube_deployment_status_replicas_unavailable > 0
+```
+
+### Deployments não atualizados
+```promql
+kube_deployment_status_observed_generation != kube_deployment_metadata_generation
+```
+
+### Proporção de réplicas disponíveis
+```promql
+kube_deployment_status_replicas_available / kube_deployment_spec_replicas
+```
+
+---
+
+## Network
+
+### Bytes recebidos por pod (rate)
+```promql
+sum(rate(container_network_receive_bytes_total[5m])) by (pod, namespace)
+```
+
+### Bytes enviados por pod (rate)
+```promql
+sum(rate(container_network_transmit_bytes_total[5m])) by (pod, namespace)
+```
+
+### Erros de rede por interface
+```promql
+sum(rate(node_network_receive_errs_total[5m])) by (instance, device)
+```
+
+### Conexões TCP por estado
+```promql
+node_netstat_Tcp_CurrEstab
+```
+
+---
+
+## Nodes
+
+### Nodes não Ready
+```promql
+kube_node_status_condition{condition="Ready",status="true"} == 0
+```
+
+### Pressão de memória
+```promql
+kube_node_status_condition{condition="MemoryPressure",status="true"} == 1
+```
+
+### Pressão de disco
+```promql
+kube_node_status_condition{condition="DiskPressure",status="true"} == 1
+```
+
+### Disco disponível por node (%)
+```promql
+(node_filesystem_avail_bytes{fstype!~"tmpfs|overlay"} / node_filesystem_size_bytes{fstype!~"tmpfs|overlay"}) * 100
+```
+
+### Load average (1 min)
+```promql
+node_load1
+```
+
+---
+
+## Cluster Overview
+
+### Total de pods Running
+```promql
+count(kube_pod_status_phase{phase="Running"})
+```
+
+### Total de namespaces
+```promql
+count(kube_namespace_created)
+```
+
+### Total de deployments
+```promql
+count(kube_deployment_created)
+```
+
+### Total de PVCs
+```promql
+count(kube_persistentvolumeclaim_info)
+```
+
+### Idade do cluster (dias)
+```promql
+(time() - min(kube_namespace_created{namespace="kube-system"})) / 86400
+```
+
+---
+
+## Victoria Metrics
+
+### Métricas sendo coletadas (por job)
+```promql
+count by (job) ({__name__!=""})
+```
+
+### Taxa de ingestão
+```promql
+sum(rate(vm_rows_inserted_total[5m]))
+```
+
+### Uso de disco do VM
+```promql
+vm_data_size_bytes
+```
+
+### Queries por segundo
+```promql
+sum(rate(vm_http_requests_total{path="/api/v1/query"}[5m]))
+```
+
+---
+
+## Dicas
+
+### Filtrar por namespace
+```promql
+# Adicione {namespace="meu-namespace"} a qualquer query
+sum(container_memory_working_set_bytes{namespace="gitlab"}) by (pod)
+```
+
+### Excluir namespaces de sistema
+```promql
+{namespace!~"kube-system|argocd|monitoring|gitlab"}
+```
+
+### Agregar por label
+```promql
+sum by (label_app) (kube_pod_info)
+```
+
+### Ordenar resultados
+```promql
+sort_desc(sum(container_memory_working_set_bytes) by (namespace))
+```
+
+### Top N
+```promql
+topk(5, sum(rate(container_cpu_usage_seconds_total[5m])) by (pod))
+```
+
+### Valor no tempo (offset)
+```promql
+# Valor de 1 hora atrás
+container_memory_working_set_bytes offset 1h
+```
+
+---
+
+## Referências
+
+- [PromQL Cheat Sheet](https://promlabs.com/promql-cheat-sheet/)
+- [Victoria Metrics MetricsQL](https://docs.victoriametrics.com/metricsql/)
+- [Grafana Dashboards](https://grafana.com/grafana/dashboards/)