README.md

How to use

Prepare

0. Edit main.tf, ready k8s.yaml config file:

provider "kubernetes" {
  config_path = "./test/k8s.yaml"  # set right k8s config file
}

1. Create PVC

cd terraform/pvc
cat <<EOF > terraform.tfvars
configs = {
  namespace = "xxxx"  # set namespace
  pvc = {
    storage_class_name = "yyyy"  # set class name
  }
}
EOF
make init
make apply

PVC should be created in a separate directory, so that PVC will not be destroyed when other resources are destroyed. (Don’t ask me why I know that PVC cannot be destroyed together)

2. write a file terraform.tfvars

configs = {
  namespace = "ahfu" # namespace on k8s
  region    = "HK"
  env       = "test"
  log = {
    format = "json"
    level  = "INFO"
    output = "stdout"
  }
  vm = {
    version = "v1.96.0"  # vectoria metrics version
  }
  self_monitor_cluster_domain = "self-monitor-cluster.my-own-test.com"
  dingtalk_webhooks = [{
    url    = "https://oapi.dingtalk.com/robot/send?access_token=${ding_talk_token}"
    secret = "${secret}"
  }]
  pvc = {
    storage_class_name = ""
    basepath           = "/vm-data/"
  }
  realtime_cluster = {
    domain             = "realtime-cluster.ahfu.com"
    storage_node_count = 3 #todo
    storage_path       = "/vm-data/realtime-cluster/sharding-"
  }
  s3 = {
    AWS_ACCESS_KEY_ID     = ""
    AWS_SECRET_ACCESS_KEY = ""
    AWS_REGION            = ""
    AWS_BUCKET            = ""
  }
}

3. config kubectl command line

kubectl config use-context ahfu

Deploy

make init
make apply

View datas from grafana

Copy realtime-cluster-ingress-ip and self-monitor-cluster-ingress-ip from terraform outputs.

sudo vi /etc/hosts

Add:

self-monitor-cluster.my-own-test.com 10.151.0.70 # change the ip to ingress ip
realtime-cluster.my-own-test.com 10.151.0.71

Then add Prometheus data source by realtime-cluster-vm-select-service and self-monitor-vm-select-service:

# like this:
http://10.43.130.240:8481/select/0/prometheus/

Finally, you can use MetricsQL to query data on grafana.