#azure (2021-04)

sorry but your question is not very clear, does your question is about single repos vs multiple repos to manage azure resources and terraform state ?

Hi there - not really - it’s more about CAF

and if anyone is following Microsoft’s best practices there.

Especially as there are terraform modules for it.

https://github.com/Azure/terraform-azurerm-caf-enterprise-scale

Essentially I’m just curious if anyone is using these modules and how it went so far.

I just discovered them when you posted them ! it’s a great source of inspiration ;)

not sure how to use it as it is by example the service_health_alerts use a provider azurecaf

terraform {
  required_providers {
    azurecaf = {
      source = "aztfmod/azurecaf"
    }
  }
}

and then has dedicated resource resource “azurecaf_name” “ag1_name” {} https://github.com/aztfmod/terraform-azurerm-caf/blob/master/modules/monitoring/service_health_alerts/main.tf

I have found azure_caf on the terraform registry and it has 129k download … https://registry.terraform.io/providers/aztfmod/azurecaf/latest

Yep

Just implemented a Proof of Concept with it.

nice !! , what was your though about it ?

I want to monitor my azure loadbalancer, vault etc with it

I was troublesome to get it working, documentation seems out of date and is inconsistent. I’m using the discovery agent as well.

But now it’s working I’m happy with it.

I’m monitoring AKS, API management and AKS Load Balancer with it

it discover Azure monitor probes ? or resources ? how does it get the prometheus port

i use prometheus azure_sd to discover resource to monitor based on tag with by example:

• node_exporter=true

• windows_exporter=true

• haproxy_exporter =true

ha I see promitor resource discovery is only for Kub !

The RD discovers Azure resources based on filters, eg:

- name: aks_vmss
  type: VirtualMachineScaleSet
  criteria:
    include:
      regions:
      - westeurope
- name: apimanagement
  type: ApiManagement
  criteria:
    include:
      regions:
      - westeurope
      - northeurope

ok I see

And then you configure Promitor to query that RD group, eg:

  - name: aks_vmss_cpuload
    description: "Percentage CPU"
    resourceType: VirtualMachineScaleSet
    azureMetricConfiguration:
      metricName: "Percentage CPU"
      aggregation:
        type: Average
    resourceDiscoveryGroups:
    - name: aks_vmss

understood

Some things I don’t like:

• By default the pods run as root (need to fix that before it is promoted beyond a poc)

• You need to explicitly specify which metric you want to scrape, it doesn’t discover all metrics available

• Log settings seem to be buggy

• Error / warnings are not descriptive

well first I will use the “Promitor Scraper” but not the discovery

Good luck! Let me know if you get stuck