OpenObserve — setup

OpenObserve — Setup de Dashboards e Alertas

Acesso

• URL: https://o2.colabh.org
• Email: ra.almeida@unesp.br
• Password: Cpps900@OpenObserve

Streams disponiveis (56 metricas)

O OTel Collector coleta automaticamente:

Node (hostmetrics)

• system_cpu_utilization — uso de CPU (0-1)
• system_memory_utilization — uso de memoria (0-1)
• system_cpu_load_average_1m/5m/15m — load average
• system_filesystem_usage — uso de disco
• system_network_io — trafego de rede (bytes)
• system_disk_io — I/O de disco

Pods/Containers (kubeletstats)

• k8s_pod_cpu_utilization — CPU por pod
• k8s_pod_memory_usage — memoria por pod
• k8s_node_cpu_utilization — CPU do node K8s
• k8s_node_memory_usage — memoria do node K8s
• container_cpu_utilization — CPU por container
• container_memory_usage — memoria por container

Labels disponiveis para filtrar: k8s_pod_name, k8s_namespace_name, k8s_node_name, k8s_deployment_name, k8s_container_name

Logs (filelog)

Logs de todos os pods sao coletados automaticamente com start_at: end (so novos).

Dashboards recomendados

Criar via UI (Dashboards → New Dashboard):

1. K3s Cluster Overview

Tab: Node

Panel	Query (PromQL)	Tipo
CPU Utilization	avg(system_cpu_utilization)	Area
Memory Utilization	avg(system_memory_utilization)	Area
Load Average	avg(system_cpu_load_average_1m), avg(system_cpu_load_average_5m), avg(system_cpu_load_average_15m)	Area
Filesystem	avg(system_filesystem_usage)	Area
Network I/O	rate(system_network_io[5m])	Area

Tab: Pods

Panel	Query (PromQL)	Tipo
Pod CPU top 10	topk(10, avg by (k8s_pod_name) (k8s_pod_cpu_utilization))	Area
Pod Memory top 10	topk(10, avg by (k8s_pod_name) (k8s_pod_memory_usage))	Area
CPU por Namespace	sum by (k8s_namespace_name) (container_cpu_utilization)	Area
Memory por Namespace	sum by (k8s_namespace_name) (container_memory_usage)	Area

2. Infrastructure

Panel	Query	Tipo
CPU por Core	system_cpu_utilization	Area
Disk I/O	rate(system_disk_io[5m])	Area
Disk Operations	rate(system_disk_operations[5m])	Area
Network Errors	rate(system_network_errors[5m])	Area

Alertas recomendados

Criar via UI (Alerts → New Alert):

Alerta	Condicao	Severidade
CPU alta	avg(system_cpu_utilization) > 0.9 por 10min	Critical
Memoria alta	avg(system_memory_utilization) > 0.9 por 10min	Critical
Disco cheio	avg(system_filesystem_usage) > 0.85	Warning
Load alto	avg(system_cpu_load_average_5m) > 4 por 5min	Warning
Pod memory alta	k8s_pod_memory_usage > 2e9 (2GB)	Warning

Arquitetura

Pods (todos os namespaces)
    |
    v
OTel Collector Agent (DaemonSet)
    |-- filelog: logs de /var/log/pods/
    |-- hostmetrics: CPU, mem, disco, rede
    |-- kubeletstats: metricas K8s
    |
    v (OTLP HTTP)
OpenObserve (:5080)
    |
    v
Dashboards + Alertas