Customised Alarm Notification in OCI

I had written a blog earlier on how to customise operation insights alarm.

In this blog we will see another sample code to customise generic alarm message using OCI function to get a short and crisp message like below

Severity : CRITICAL

Resource : <resourceDisplayName>

Resource ID : ocid1.instance.oc1…..ancd

Alarm Query and Value: {‘CpuUtilization[1m].mean()’: ‘70.08’}

import io
import oci
import json


def ons_publish(**kwargs):
    try:
        signer = oci.auth.signers.get_resource_principals_signer()
        ons_client = oci.ons.NotificationDataPlaneClient({},signer=signer)

        topic_id = kwargs.get('topic_id')
        resource_display_name_func = kwargs.get('resource_display_name')
        severity_func = kwargs.get('severity')
        metric_value_func = kwargs.get('metric_value')
        title_func = kwargs.get('title')
        resource_id_func = kwargs.get('resource_id')

        alarm_body = f'Severity : {severity_func}\nResource : {resource_display_name_func}\nResource ID : {resource_id_func}\nAlarm Query and Value: {metric_value_func}'

        print("Publishing message to ONS topic")
        ons_client.publish_message(
            topic_id=topic_id,
            message_details=oci.ons.models.MessageDetails(
                body=alarm_body,
                title=title_func))
    except Exception as ons_exception:
        print(ons_exception)


def handler(ctx, data: io.BytesIO = None):
    cfg = dict(ctx.Config())

    # fetch details from function config
    topic_id = cfg['topic_id']

    try:
        body = json.loads(data.getvalue())
        alarm_type = body.get("type")
        alarm_status = body['alarmMetaData'][0]['status']

        #Look only for OK_TO_FIRING  and REPEAT type alarm. FIRING_TO_OK is ignored
        if alarm_type in ["OK_TO_FIRING",'REPEAT'] and alarm_status == "FIRING":

            resource_id = body["alarmMetaData"][0]["dimensions"][0]["resourceId"]
            severity = body.get("severity")
            resource_display_name = body["alarmMetaData"][0]["dimensions"][0]["resourceDisplayName"]
            metric_value = body["alarmMetaData"][0]["metricValues"][0]
            title = body.get("title")

            ons_publish(resource_display_name=resource_display_name, resource_id=resource_id, topic_id=topic_id,
                        severity=severity,
                        metric_value=metric_value, title=title)
        else:
            print("Alarm type is not in OK_TO_FIRING  or REPEAT")
    except (Exception, ValueError) as ex:
        print(ex)

Deploy the function using Code Editor or Cloud Shell . Set the topic_id in function configuration once deployed.

To test the function is working fine create a sample alarm.json file in /tmp directory

{
  "dedupeKey": "5db34-4f299401",
  "title": "Testing Alarm",
  "type": "REPEAT",
  "severity": "CRITICAL",
  "timestampEpochMillis": 1705743900000,
  "timestamp": "2024-01-20T09:45:00Z",
  "alarmMetaData": [
    {
      "id": "alarm ocid",
      "status": "FIRING",
      "severity": "CRITICAL",
      "namespace": "oci_computeagent",
      "query": "CpuUtilization[1m].mean() > 60",
      "totalMetricsFiring": 1,
      "dimensions": [
        {
          "instancePoolId": "Default",
          "resourceDisplayName": "computevmname",
          "faultDomain": "FAULT-DOMAIN-2",
          "resourceId": "resource ocid",
          "availabilityDomain": "NoEK:EU-FRANKFURT-1-AD-1",
          "imageId": "image ocid",
          "shape": "VM.Standard.E4.Flex",
          "dedicatedVmHostId": "DefaultVmHostId",
          "region": "eu-frankfurt-1"
        }
      ],
      "alarmUrl": "https://cloud.oracle.com/monitoring/alarms",
      "alarmSummary": "Alarm \"Testing Alarm\" is in a \"FIRING\" state; because the resources with dimensions listed below meet the trigger rule: \"CpuUtilization[1m].mean() > 60\", with a trigger delay of 1 minute",
      "metricValues": [
        {
          "CpuUtilization[1m].mean()": "70.08"
        }
      ]
    }
  ],
  "notificationType": "Split messages per metric stream",
  "version": 1.5
}

In the Cloud Shell or Code Editor invoke the function using the command

cat /tmp/alarm.json | fn invoke <application> <functionname>