Merge branch 'keda-upgrade-1' into 'master'

Keda upgrade from 1.5 to 2

See merge request !430

charts/airflow/helm-config-dp.yaml

@@ -71,7 +71,7 @@ pgbouncer:
 # Specify KEDA configuration
 #
 keda:
-  version_2_enabled: false
+  version_2_enabled: true
 
 ################################################################################
 # Specify the airflow configuration

charts/airflow/templates/airflow-log-processor-deployment.yaml

-{{- $isEnabled := .Values.airflow.isDataPartitionDeployment | default false -}}
+{{- $isEnabled := .Values.keda.version_2_enabled | default false -}}
 apiVersion: apps/v1
 kind: Deployment
 metadata:
@@ -41,20 +41,6 @@ spec:
               secretKeyRef:
                 name: airflow
                 key: storage-connection
-        
----
-{{- if $isEnabled }}
-apiVersion: keda.sh/v1alpha1
-{{- else }}
-apiVersion: keda.k8s.io/v1alpha1
-{{- end }}
-kind: TriggerAuthentication
-metadata:
-  name: azure-queue-auth
-spec:
-  podIdentity:
-    provider: azure
-
 ---
 kind: ScaledObject
 metadata:
@@ -73,9 +59,10 @@ spec:
   triggers:
   - type: azure-queue
     metadata:
-      # Required
       queueName: airflowlogqueue
-      # Optional
       queueLength: "5" # default 5
-    authenticationRef:
-        name: azure-queue-auth # authenticationRef would need either podIdentity or define a connection parameter
\ No newline at end of file
+      {{- if $isEnabled }}
+      connectionFromEnv: AzureWebJobsStorage
+      {{- else }}
+      connection: AzureWebJobsStorage
+      {{- end }}
\ No newline at end of file

docs/keda-upgrade.md

+## Upgade keda version from 1.5 to 2.x
+
+We've used feature flag to upgrade keda version from 1.5 to 2.x osdu infra on azure.
+A user can follow following steps to upgrade keda version - 
+
+1. We've set keda_v2_enabled field in feature_flag variable in service resource to false.
+It represents that currently osdu infra is currently using 1.5 version.
+2. If using flux, reduce replica count to 0 for flux deployment until step 4. (use command - kubectl edit deployment flux -n flux)
+3. Either run it or follow the steps in this script - infra/scripts/keda_upgrade_and_host_encryption.sh. 
+   This script requires some inputs. Feel free to edit the script as per requirement.
+4. Override keda_v2_enabled field to true.
+
+
+Service deployment steps for manual users - 
+1. We've added a variable named (keda.version_2_enabled: false) in Values.yaml for following files in helm-charts-azure repo-
+    - osdu-airflow/values.yaml
+    - osdu-azure/osdu-core_services/values.yaml 
+    - osdu-azure/osdu-ingest_enrich/values.yaml 
+    
+2. Override (keda.version_2_enabled: false) value to true.
+3. Make the deployment.
+
+Service deployment steps for automated pipeline users - 
+1. We've added a variable named (keda.version_2_enabled: false) in Values.yaml in indexer-queue repo.
+2. Override (keda.version_2_enabled: false) value here to true.
+3. Make the deployment
+   
+    

infra/scripts/keda_upgrade_and_host_encryption.sh

+#!/usr/bin/env bash
+
+read -p "ARM_TENANT_ID: " tenantId
+read -p "ARM_CLIENT_SECRET: " servicePrincipalKey
+read -p "ARM_CLIENT_ID: " servicePrincipalId
+read -p "ARM_SUBSCRIPTION_ID: " ARM_SUBSCRIPTION_ID
+read -p "Remote State Account(Service Resource): " TF_VAR_remote_state_account
+read -p "Remote State Container(Service Resource): " TF_VAR_remote_state_container
+read -p "Azure Resource Group with AKS Cluster(Service Resource): " myResourceGroup
+read -p "AKS Cluster(Service Resource): " myAKSCluster
+read -p "ARM Access Key: " storageAccountArmAccessKey
+read -p "Terraform workspace for central resources: " centralResourceTerraformWorkspace
+read -p "Gitops Branch: " gitopsBranch
+read -p "Gitops ssh url: " gitopsSshUrl
+read -p "Gitops path: " gitopsPath
+read -p "Gitops ssh public key file: " sshPublicKeyFile
+read -p "Gitops ssh key file: " gitopsSshKeyFile
+read -p "Terraform workspace name: " tfWorkspaceName
+read -p "Resource Group Location: " resourceGroupLocation
+read -p "Git Repo link with feature flag for keda v2 present.(Script assumes that repo name is infra-azure-provisioning. You can edit it if its something else): " gitRepo
+read -p "Repo branch name with feature flag for keda v2 present. This branch should be in sync with your current infra setup: " repoBranch
+
+read -p "Automated Pipeline deployment (only yes is accepted as true. Any other input will be treated as manual deployment): " automatedDeployment
+
+read -p "Application namespace (Leave empty for default value - osdu. You'll need to edit this script if applications are running in different namespaces): " namespace
+
+export TF_VERSION="0.14.4"
+export TF_VAR_gitops_ssh_url="$gitopsSshUrl"
+export TF_VAR_gitops_path="$gitopsPath"
+export TF_VAR_gitops_branch="$gitopsBranch"
+export TF_VAR_ssh_public_key_file="$sshPublicKeyFile"
+export TF_VAR_gitops_ssh_key_file="$gitopsSshKeyFile"
+export TF_WORKSPACE="$tfWorkspaceName"
+export TF_VAR_remote_state_account="$TF_VAR_remote_state_account"
+export TF_VAR_remote_state_container="$TF_VAR_remote_state_container"
+export TF_VAR_resource_group_location="$resourceGroupLocation"
+
+function enableFlux {
+  echo "Enabling flux"
+  kubectl patch deployment flux -p '{"spec":{"replicas": 1}}' -n flux
+}
+
+function disableFlux {
+  echo "Disabling flux"
+  kubectl patch deployment flux -p '{"spec":{"replicas": 0}}' -n flux
+}
+
+trap enableFlux EXIT
+
+echo "Script Started"
+if [ ! "$namespace" ]; then
+  echo "Assigning default namespace: osdu"
+  namespace=osdu
+fi
+
+echo "Namespace: $namespace"
+
+function terraformVersionCheck() {
+  if [[ $(which terraform) && $(terraform --version | head -n1 | cut -d" " -f2 | cut -c 2\-) == $TF_VERSION ]]; then
+    echo "Terraform version check completed"
+  else
+    TF_ZIP_TARGET="https://releases.hashicorp.com/terraform/$TF_VERSION/terraform_${TF_VERSION}_linux_amd64.zip"
+    echo "Info: installing $TF_VERSION, target: $TF_ZIP_TARGET"
+
+    wget $TF_ZIP_TARGET -q
+    unzip -q "terraform_${TF_VERSION}_linux_amd64.zip"
+    sudo mv terraform /usr/local/bin
+    rm *.zip
+  fi
+
+  terraform -version
+  # Assert that jq is available, and install if it's not
+  command -v jq >/dev/null 2>&1 || {
+    echo >&2 "Installing jq"
+    sudo apt install -y jq
+  }
+}
+#terraformVersionCheck
+
+#echo "Terraform verion check complete"
+
+echo "Logging in to az cli"
+az login --service-principal -u "$servicePrincipalId" --password="$servicePrincipalKey" --tenant "$tenantId"
+
+echo "Login successful"
+
+az aks get-credentials --resource-group="$myResourceGroup" --name="$myAKSCluster"
+
+#echo "Credential fetch successful"
+
+az account set --subscription "$ARM_SUBSCRIPTION_ID"
+
+echo "Subscription set to $ARM_SUBSCRIPTION_ID"
+
+export ARM_TENANT_ID=$tenantId
+export ARM_CLIENT_SECRET=$servicePrincipalKey
+export ARM_CLIENT_ID=$servicePrincipalId
+export ARM_SUBSCRIPTION_ID="$ARM_SUBSCRIPTION_ID"
+export TF_VAR_central_resources_workspace_name="$centralResourceTerraformWorkspace"
+
+export ARM_ACCESS_KEY="$storageAccountArmAccessKey"
+
+disableFlux
+
+echo "Deleting existing scaled objects"
+NEXT_WAIT_TIME=0
+until [ $NEXT_WAIT_TIME -eq 5 ] || timeout 3 kubectl delete scaledobjects.keda.k8s.io --all; do
+    sleep $(( NEXT_WAIT_TIME++ ))
+done
+[ "$NEXT_WAIT_TIME" -lt 5 ]
+
+
+echo "Deleting existing trigger authentications"
+NEXT_WAIT_TIME=0
+until [ $NEXT_WAIT_TIME -eq 5 ] || timeout 3 kubectl delete triggerauthentications.keda.k8s.io --all; do
+    sleep $(( NEXT_WAIT_TIME++ ))
+done
+[ "$NEXT_WAIT_TIME" -lt 5 ]
+
+#set -euo pipefail
+
+echo "Uninstalling Keda v1"
+helm uninstall -n keda keda
+
+echo "Deleting Keda v1 CRDs"
+
+NEXT_WAIT_TIME=0
+until [ $NEXT_WAIT_TIME -eq 5 ] || timeout 3 kubectl delete crd scaledobjects.keda.k8s.io --ignore-not-found=true; do
+    kubectl patch crd/scaledobjects.keda.k8s.io -p '{"metadata":{"finalizers":[]}}' --type=merge
+    sleep $(( NEXT_WAIT_TIME++ ))
+done
+[ "$NEXT_WAIT_TIME" -lt 5 ]
+
+NEXT_WAIT_TIME=0
+until [ $NEXT_WAIT_TIME -eq 5 ] || timeout 3 kubectl delete crd triggerauthentications.keda.k8s.io --ignore-not-found=true; do
+    kubectl patch crd/triggerauthentications.keda.k8s.io -p '{"metadata":{"finalizers":[]}}' --type=merge
+    sleep $(( NEXT_WAIT_TIME++ ))
+done
+[ "$NEXT_WAIT_TIME" -lt 5 ]
+
+set -euo pipefail
+
+echo "terraform applying with keda version updated to 2.2.0"
+
+git clone --branch $repoBranch $gitRepo
+
+# Assuming repo name is infra-azure-provisioning. If its not, edit the repo name in below command and cd into service_resources.
+cd infra-azure-provisioning/infra/templates/osdu-r3-mvp/service_resources
+echo "keda_v2_enabled = true" >> override.tfvars
+terraform init -upgrade -backend-config "storage_account_name=$TF_VAR_remote_state_account" -backend-config "container_name=$TF_VAR_remote_state_container"
+
+terraform state rm helm_release.keda
+
+echo "Terraform applying"
+terraform apply -var-file override.tfvars
\ No newline at end of file

infra/templates/osdu-r3-mvp/service_resources/helm_keda.tf

@@ -20,7 +20,7 @@ locals {
   helm_keda_name    = "keda"
   helm_keda_ns      = "keda"
   helm_keda_repo    = "https://kedacore.github.io/charts"
-  helm_keda_version = "1.5.0"
+  helm_keda_version = var.keda_v2_enabled == true ? "2.2.0" : "1.5.0"
 }
 
 resource "kubernetes_namespace" "keda" {

infra/templates/osdu-r3-mvp/service_resources/override.tfvars

@@ -28,6 +28,8 @@ feature_flag = {
   autoscaling    = false
 }
 
+keda_v2_enabled = false
+
 prefix = "osdu-mvp"
 
 resource_tags = {

infra/templates/osdu-r3-mvp/service_resources/terraform.tfvars

@@ -73,6 +73,9 @@ feature_flag = {
   autoscaling    = false
 }
 
+# Feature flag for keda v2
+keda_v2_enabled = false
+
 # cosmos DB SQL collections
 cosmos_sql_collections = [
   {

infra/templates/osdu-r3-mvp/service_resources/variables.tf

@@ -35,7 +35,6 @@ variable "feature_flag" {
     flux           = bool
     sa_lock        = bool
     autoscaling    = bool
-
   })
   default = {
     osdu_namespace = true
@@ -45,6 +44,12 @@ variable "feature_flag" {
   }
 }
 
+variable "keda_v2_enabled" {
+  description = "Feature flag for kuda v2. Needs to be true if keda v2 is installed else false"
+  type        = bool
+  default     = false
+}
+
 variable "randomization_level" {
   description = "Number of additional random characters to include in resource names to insulate against unexpected resource name collisions."
   type        = number