Deploying VictoriaMetrics in Cluster Mode (by Components)

VictoriaMetrics in Cluster Mode consists of several components that can be deployed separately for horizontal scaling and high availability.

Components in Cluster Mode:

Component	Description
VMInsert	Receives data from Prometheus/Graphite/OpenTSDB and forwards it to VMStorage.
VMSelect	Executes queries against VMStorage and connects to Grafana and PromQL.
VMStorage	The main data storage, works with SSTable (similar to TSDB in Prometheus).
VMAgent	An alternative to Prometheus, collects metrics from endpoints.
VMAlert	Sends alerts, similar to Alertmanager.

1. Running VictoriaMetrics Cluster with Docker Compose

This setup deploys all components with the necessary port mappings.

docker-compose.yml File

version: '3.7'

services:
  vmstorage:
    image: victoriametrics/vmstorage
    container_name: vmstorage
    restart: always
    ports:
      - "8400:8400"
    volumes:
      - ./data:/storage
    command:
      - "-storageDataPath=/storage"
      - "-retentionPeriod=30"  # Store metrics for 30 days

  vminsert:
    image: victoriametrics/vminsert
    container_name: vminsert
    restart: always
    ports:
      - "8480:8480"
    command:
      - "-storageNode=vmstorage:8400"

  vmselect:
    image: victoriametrics/vmselect
    container_name: vmselect
    restart: always
    ports:
      - "8428:8428"
    command:
      - "-storageNode=vmstorage:8400"

  vmagent:
    image: victoriametrics/vmagent
    container_name: vmagent
    restart: always
    ports:
      - "8429:8429"
    command:
      - "-remoteWrite.url=http://vminsert:8480/insert/0/prometheus"

  vmalert:
    image: victoriametrics/vmalert
    container_name: vmalert
    restart: always
    ports:
      - "8880:8880"
    command:
      - "-datasource.url=http://vmselect:8428"
      - "-remoteWrite.url=http://vminsert:8480"

Starting the Cluster

docker-compose up -d

Now the components are accessible at:

• VMInsert (metric ingestion): http://localhost:8480/insert/0/prometheus
• VMSelect (queries for Grafana): http://localhost:8428
• VMStorage (main storage): http://localhost:8400
• VMAgent (metric collector): http://localhost:8429
• VMAlert (alerting): http://localhost:8880

2. Configuring Grafana with VictoriaMetrics

Adding Data Source in Grafana

1. Open Grafana (http://localhost:3000).
2. Navigate to Configuration → Data Sources.
3. Click Add Data Source.
4. Select Prometheus.
5. In the URL field, enter:

http://localhost:8428

1. Click Save & Test.

Now Grafana can query VictoriaMetrics using PromQL.

3. Example Metrics and Queries

Example PromQL Queries for Grafana

• CPU Usage:

rate(node_cpu_seconds_total[5m])

• Memory Usage:

node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes * 100

• Number of Active Pods in Kubernetes:

count(kube_pod_info)

Conclusion

✅ Successfully deployed VictoriaMetrics in Cluster Mode
✅ Connected Grafana
✅ Ready to collect and analyze metrics

The system is now scalable, resilient, and capable of processing metrics efficiently!