Module fixes and improvements

terraform/files/node-problem-detector/values.yaml

@@ -457,8 +457,12 @@ settings:
 
         cp ./config/*.py /host/tmp
 
-        chroot /host bash -c "chmod +x /tmp/check_gpu_setup.py"
-        chroot /host bash -c "/tmp/check_gpu_setup.py --slurm --$healthcheck_type > /tmp/oke-latest-${healthcheck_type}-healthcheck.log 2>&1"
+        if [ ! -e "/host/tmp/uv" ]; then
+            cp /usr/bin/uv /host/tmp/uv
+            chmod +x /host/tmp/uv
+        fi
+
+        EXIT_CODE=$(nsenter -n/host/proc/1/ns/net chroot /host bash -c "/tmp/uv run /tmp/check_gpu_setup.py --dry-run --$healthcheck_type > /tmp/oke-latest-${healthcheck_type}-healthcheck.log 2>&1")
 
         ERROR_MSG=$(cat "$healtcheck_log_file" | grep "Healthcheck::")
         UNKNOWN_MSG=$(cat "$healtcheck_log_file" | grep -o 'Skipping.*')
@@ -467,7 +471,7 @@ settings:
         then
             echo "${ERROR_MSG#* }"
             exit $NONOK
-        elif [ "$ERROR_MSG" == "" ] && [ "$UNKNOWN_MSG" != "" ]
+        elif [ ! "${EXIT_CODE}" ] && [ "$UNKNOWN_MSG" != "" ]
         then
             echo $UNKNOWN_MSG
             exit $UNKNOWN
@@ -521,7 +525,7 @@ logDir:
 
 image:
   repository: iad.ocir.io/idxzjcdglx2s/oke-npd
-  tag: v1.34.0-1
+  tag: v1.34.0-2
   # image.digest -- the image digest. If given it takes precedence over a given tag.
   digest: ""
   pullPolicy: IfNotPresent

terraform/oke-cluster.tf

@@ -36,7 +36,7 @@ locals {
       bastion  = var.create_bastion ? { create = "auto" } : { create = "never"}
       operator = var.create_operator ? { create = "auto" } : { create = "never"}
       int_lb   = { create = "auto" }
-      pub_lb   = alltrue([!var.create_vcn, var.pub_lb_sn_id == null, var.pub_lb_sn_cidr == null]) ? { create = "never" } : { create = "auto"}
+      pub_lb   = { create = "auto" }
       cp       = { create = "auto" }
       workers  = { create = "auto" }
       pods     = { create = "auto" }
@@ -50,89 +50,97 @@ locals {
     {
       bastion = merge(
         var.create_bastion ? { create = "auto" } : { create = "never" },
-        (var.create_vcn && var.bastion_sn_cidr == null) || (!var.create_vcn && var.bastion_sn_id == null) ?
+        (var.create_vcn && var.bastion_sn_cidr == null) || (!var.create_vcn && !var.custom_subnet_ids) ?
         { newbits = 13, netnum = 1 } : {},
         var.create_vcn && var.bastion_sn_cidr != null ?
         { cidr = var.bastion_sn_cidr } : {},
-        var.vcn_id != null && var.bastion_sn_id != null ?
-        { id = var.bastion_sn_id } : {}
+        !var.create_vcn && var.custom_subnet_ids ?
+        { id = var.bastion_sn_id, create = "never" } : {},
+        lookup(var.subnet_advanced_attrs, "bastion", {})
       )
-      operator = var.create_operator ? merge(
+      operator = merge(
         var.create_operator ? { create = "auto" } : { create = "never" },
-        (var.create_vcn && var.operator_sn_cidr == null) || (!var.create_vcn && var.operator_sn_id == null) ?
+        (var.create_vcn && var.operator_sn_cidr == null) || (!var.create_vcn && !var.custom_subnet_ids) ?
         { newbits = 13, netnum = 2 } : {},
         var.create_vcn && var.operator_sn_cidr != null ?
         { cidr = var.operator_sn_cidr } : {},
-        var.vcn_id != null && var.operator_sn_id != null ?
-        { id = var.operator_sn_id } : {}
-      ) : { create = "never" }
+        !var.create_vcn && var.custom_subnet_ids ?
+        { id = var.operator_sn_id, create = "never" } : {},
+        lookup(var.subnet_advanced_attrs, "operator", {})
+      )
       int_lb = merge(
         { create = "auto" },
-        (var.create_vcn && var.int_lb_sn_cidr == null) || (!var.create_vcn && var.int_lb_sn_id == null) ?
+        (var.create_vcn && var.int_lb_sn_cidr == null) || (!var.create_vcn && !var.custom_subnet_ids) ?
         { newbits = 11, netnum = 1 } : {},
         var.create_vcn && var.int_lb_sn_cidr != null ?
         { cidr = var.int_lb_sn_cidr } : {},
-        var.vcn_id != null && var.int_lb_sn_id != null ?
-        { id = var.int_lb_sn_id } : {}
+        !var.create_vcn && var.custom_subnet_ids ?
+        { id = var.int_lb_sn_id, create = "never" } : {},
+        lookup(var.subnet_advanced_attrs, "int_lb", {})
       )
       pub_lb = merge(
         { create = "auto" },
-        (var.create_vcn && var.pub_lb_sn_cidr == null) || (!var.create_vcn && var.pub_lb_sn_id == null) ?
+        (var.create_vcn && var.pub_lb_sn_cidr == null) || (!var.create_vcn && !var.custom_subnet_ids) ?
         { newbits = 11, netnum = 2 } : {},
         var.create_vcn && var.pub_lb_sn_cidr != null ?
         { cidr = var.pub_lb_sn_cidr } : {},
-        var.vcn_id != null && var.pub_lb_sn_id != null ?
-        { id = var.pub_lb_sn_id } : {},
-        alltrue([!var.create_vcn, var.pub_lb_sn_id == null, var.pub_lb_sn_cidr == null]) ? { create = "never" } : {}
+        !var.create_vcn && var.custom_subnet_ids ?
+        { id = var.pub_lb_sn_id, create = "never" } : {},
+        lookup(var.subnet_advanced_attrs, "pub_lb", {})
       )
       cp = merge(
         { create = "auto" },
-        (var.create_vcn && var.cp_sn_cidr == null) || (!var.create_vcn && var.cp_sn_id == null) ?
+        (var.create_vcn && var.cp_sn_cidr == null) || (!var.create_vcn && !var.custom_subnet_ids) ?
         { newbits = 13, netnum = 0 } : {},
         var.create_vcn && var.cp_sn_cidr != null ?
         { cidr = var.cp_sn_cidr } : {},
-        var.vcn_id != null && var.cp_sn_id != null ?
-        { id = var.cp_sn_id } : {}
+        !var.create_vcn && var.custom_subnet_ids ?
+        { id = var.cp_sn_id, create = "never" } : {},
+        lookup(var.subnet_advanced_attrs, "cp", {})
       )
       workers = merge(
         { create = "auto" },
-        (var.create_vcn && var.workers_sn_cidr == null) || (!var.create_vcn && var.workers_sn_id == null) ?
+        (var.create_vcn && var.workers_sn_cidr == null) || (!var.create_vcn && !var.custom_subnet_ids) ?
         { newbits = 4, netnum = 2 } : {},
         var.create_vcn && var.workers_sn_cidr != null ?
         { cidr = var.workers_sn_cidr } : {},
-        var.vcn_id != null && var.workers_sn_id != null ?
-        { id = var.workers_sn_id } : {}
+        !var.create_vcn && var.custom_subnet_ids ?
+        { id = var.workers_sn_id, create = "never" } : {},
+        lookup(var.subnet_advanced_attrs, "workers", {})
       )
       pods = merge(
         { create = "auto" },
-        (var.create_vcn && var.pods_sn_cidr == null) || (!var.create_vcn && var.pods_sn_id == null) ?
+        (var.create_vcn && var.pods_sn_cidr == null) || (!var.create_vcn && !var.custom_subnet_ids) ?
         { newbits = 2, netnum = 2 } : {},
         var.create_vcn && var.pods_sn_cidr != null ?
         { cidr = var.pods_sn_cidr } : {},
-        var.vcn_id != null && var.pods_sn_id != null ?
-        { id = var.pods_sn_id } : {}
+        !var.create_vcn && var.custom_subnet_ids ?
+        { id = var.pods_sn_id, create = "never" } : {},
+        lookup(var.subnet_advanced_attrs, "pods", {})
       )
     },
     var.create_fss ? {
       fss = merge(
         { create = "always" },
-        (var.create_vcn && var.fss_sn_cidr == null) || (!var.create_vcn && var.fss_sn_id == null) ?
+        (var.create_vcn && var.fss_sn_cidr == null) || (!var.create_vcn && !var.custom_subnet_ids) ?
         { newbits = 11, netnum = 3 } : {},
         var.create_vcn && var.fss_sn_cidr != null ?
         { cidr = var.fss_sn_cidr } : {},
-        var.vcn_id != null && var.fss_sn_id != null ?
-        { id = var.fss_sn_id } : {}
+        !var.create_vcn && var.custom_subnet_ids ?
+        { id = var.fss_sn_id, create = "never" } : {},
+        lookup(var.subnet_advanced_attrs, "fss", {})
       )
     } : {},
     var.create_lustre ? {
       lustre = merge(
         { create = "always" },
-        (var.create_vcn && var.lustre_sn_cidr == null) || (!var.create_vcn && var.lustre_sn_id == null) ?
+        (var.create_vcn && var.lustre_sn_cidr == null) || (!var.create_vcn && !var.custom_subnet_ids) ?
         { newbits = 7, netnum = 1 } : {},
         var.create_vcn && var.lustre_sn_cidr != null ?
         { cidr = var.lustre_sn_cidr } : {},
-        var.vcn_id != null && var.lustre_sn_id != null ?
-        { id = var.lustre_sn_id } : {}
+        !var.create_vcn && var.custom_subnet_ids ?
+        { id = var.lustre_sn_id, create = "never" } : {},
+        lookup(var.subnet_advanced_attrs, "lustre", {})
       )
     } : {}
   )

terraform/schema.yaml

@@ -43,6 +43,7 @@ variableGroups:
       # Monitoring
       - wildcard_dns_domain
       - monitoring_advanced_options
+      - subnet_advanced_attrs
 
   - title: "Identity"
     variables:

terraform/variables.tf

@@ -69,6 +69,10 @@ variable "workers_sn_id" { default = null }
 variable "pods_sn_id" { default = null }
 variable "fss_sn_id" { default = null }
 variable "lustre_sn_id" { default = null }
+variable "subnet_advanced_attrs" { 
+  default = {}
+  type    = any
+}
 variable "networking_advanced_options" {
   type    = bool
   default = false

terraform/via-operator-helm-deployments.tf

@@ -24,16 +24,16 @@ module "nginx" {
   post_deployment_commands = flatten([
     "cat <<'EOF' | kubectl apply -f -",
     ( var.use_lets_encrypt_prod_endpoint == true ? 
-      split("\n", file("${path.root}/files/cert-manager/cluster-issuer-prod.yaml")) :
-      split("\n", file("${path.root}/files/cert-manager/cluster-issuer-staging.yaml"))
+      split("\n", file("${path.module}/files/cert-manager/cluster-issuer-prod.yaml")) :
+      split("\n", file("${path.module}/files/cert-manager/cluster-issuer-staging.yaml"))
     ),
     "EOF",
     "sleep 60" #wait for the LB to be provisioned
   ])
   deployment_extra_args = ["--wait"]
 
   helm_template_values_override = templatefile(
-    "${path.root}/files/nginx-ingress/values.yaml.tpl",
+    "${path.module}/files/nginx-ingress/values.yaml.tpl",
     {
       min_bw    = 10,
       max_bw    = 100,
@@ -98,7 +98,7 @@ module "kube_prometheus_stack" {
 
   post_deployment_commands = []
 
-  helm_template_values_override = templatefile("./files/kube-prometheus/values.yaml.tftpl", { preferred_kubernetes_services = var.preferred_kubernetes_services})
+  helm_template_values_override = templatefile("${path.module}/files/kube-prometheus/values.yaml.tftpl", { preferred_kubernetes_services = var.preferred_kubernetes_services})
 
   helm_user_values_override = yamlencode(
     {
@@ -143,7 +143,7 @@ module "node_problem_detector" {
   deployment_extra_args    = ["--force", "--dependency-update", "--history-max 1"]
   post_deployment_commands = []
 
-  helm_template_values_override = file("${path.root}/files/node-problem-detector/values.yaml")
+  helm_template_values_override = file("${path.module}/files/node-problem-detector/values.yaml")
   helm_user_values_override     = ""
 
   depends_on = [module.kube_prometheus_stack]
@@ -162,7 +162,7 @@ module "nvidia_dcgm_exporter" {
 
   deployment_name    = "dcgm-exporter"
   namespace          = var.monitoring_namespace
-  helm_chart_path    = "${path.root}/files/nvidia-dcgm-exporter"
+  helm_chart_path    = "${path.module}/files/nvidia-dcgm-exporter"
   helm_chart_version = var.dcgm_exporter_chart_version
 
   pre_deployment_commands = [
@@ -172,7 +172,7 @@ module "nvidia_dcgm_exporter" {
   deployment_extra_args    = ["--force", "--dependency-update", "--history-max 1"]
   post_deployment_commands = []
 
-  helm_template_values_override = file("${path.root}/files/nvidia-dcgm-exporter/oke-values.yaml")
+  helm_template_values_override = file("${path.module}/files/nvidia-dcgm-exporter/oke-values.yaml")
   helm_user_values_override     = ""
 
   depends_on = [module.kube_prometheus_stack]
@@ -202,7 +202,7 @@ module "amd_device_metrics_exporter" {
   deployment_extra_args    = ["--force", "--dependency-update", "--history-max 1"]
   post_deployment_commands = []
 
-  helm_template_values_override = file("${path.root}/files/amd-device-metrics-exporter/values.yaml")
+  helm_template_values_override = file("${path.module}/files/amd-device-metrics-exporter/values.yaml")
   helm_user_values_override     = ""
 
   depends_on = [module.kube_prometheus_stack]
@@ -232,7 +232,7 @@ module "lustre_client" {
   post_deployment_commands = var.create_lustre_pv ? flatten([
     "cat <<'EOF' | kubectl apply -f -",
     split("\n", templatefile(
-      "${path.root}/files/lustre/lustre-pv.yaml.tpl",
+      "${path.module}/files/lustre/lustre-pv.yaml.tpl",
       {
         lustre_storage_size = floor(var.lustre_size_in_tb),
         lustre_ip           = one(oci_lustre_file_storage_lustre_file_system.lustre.*.management_service_address),
@@ -261,7 +261,7 @@ module "oke-ons-webhook" {
 
   deployment_name    = "oke-ons-webhook"
   namespace          = var.monitoring_namespace
-  helm_chart_path    = "${path.root}/files/oke-ons-webhook"
+  helm_chart_path    = "${path.module}/files/oke-ons-webhook"
   helm_chart_version = var.oke_ons_webhook_chart_version
 
   pre_deployment_commands = [

terraform/via-provider-amd-device-metrics-exporter.tf

@@ -9,7 +9,7 @@ resource "helm_release" "amd_device_metrics_exporter" {
   chart             = "device-metrics-exporter-charts"
   repository        = "https://rocm.github.io/device-metrics-exporter"
   version           = var.amd_device_metrics_exporter_chart_version
-  values            = ["${file("./files/amd-device-metrics-exporter/values.yaml")}"]
+  values            = ["${file("${path.module}/files/amd-device-metrics-exporter/values.yaml")}"]
   create_namespace  = false
   recreate_pods     = true
   force_update      = true

terraform/via-provider-kube-prometheus-stack.tf

@@ -14,7 +14,7 @@ resource "helm_release" "prometheus" {
   chart             = "kube-prometheus-stack"
   repository        = "https://prometheus-community.github.io/helm-charts"
   version           = var.prometheus_stack_chart_version
-  values            = ["${templatefile("./files/kube-prometheus/values.yaml.tftpl", { preferred_kubernetes_services = var.preferred_kubernetes_services})}"]
+  values            = ["${templatefile("${path.module}/files/kube-prometheus/values.yaml.tftpl", { preferred_kubernetes_services = var.preferred_kubernetes_services})}"]
   create_namespace  = true
   recreate_pods     = false
   force_update      = true

terraform/via-provider-lustre-client.tf

@@ -31,7 +31,7 @@ resource "kubectl_manifest" "lustre_pv" {
   ]
 
   yaml_body = templatefile(
-    "${path.root}/files/lustre/lustre-pv.yaml.tpl",
+    "${path.module}/files/lustre/lustre-pv.yaml.tpl",
     {
       lustre_storage_size = floor(var.lustre_size_in_tb),
       lustre_ip           = one(oci_lustre_file_storage_lustre_file_system.lustre.*.management_service_address),

terraform/via-provider-nginx.tf

@@ -15,7 +15,7 @@ resource "helm_release" "nginx" {
   version    = var.nginx_chart_version
   values = [
     templatefile(
-      "${path.root}/files/nginx-ingress/values.yaml.tpl",
+      "${path.module}/files/nginx-ingress/values.yaml.tpl",
       {
         min_bw    = 10,
         max_bw    = 100,
@@ -47,8 +47,8 @@ resource "kubectl_manifest" "cluster_issuer" {
   ]
 
   yaml_body = (var.use_lets_encrypt_prod_endpoint ?
-    file("${path.root}/files/cert-manager/cluster-issuer-prod.yaml") :
-    file("${path.root}/files/cert-manager/cluster-issuer-staging.yaml")
+    file("${path.module}/files/cert-manager/cluster-issuer-prod.yaml") :
+    file("${path.module}/files/cert-manager/cluster-issuer-staging.yaml")
   )
 }
 

terraform/via-provider-node-problem-detector.tf

@@ -9,7 +9,7 @@ resource "helm_release" "node-problem_detector" {
   chart             = "node-problem-detector"
   repository        = "oci://ghcr.io/deliveryhero/helm-charts"
   version           = var.node_problem_detector_chart_version
-  values            = ["${file("./files/node-problem-detector/values.yaml")}"]
+  values            = ["${file("${path.module}/files/node-problem-detector/values.yaml")}"]
   create_namespace  = true
   recreate_pods     = true
   force_update      = true

terraform/via-provider-nvidia-dcgm-exporter.tf

@@ -6,9 +6,9 @@ resource "helm_release" "nvidia_dcgm_exporter" {
   depends_on        = [helm_release.prometheus]
   namespace         = var.monitoring_namespace
   name              = "dcgm-exporter"
-  chart             = "${path.root}/files/nvidia-dcgm-exporter"
+  chart             = "${path.module}/files/nvidia-dcgm-exporter"
   version           = var.dcgm_exporter_chart_version
-  values            = ["${file("./files/nvidia-dcgm-exporter/oke-values.yaml")}"]
+  values            = ["${file("${path.module}/files/nvidia-dcgm-exporter/oke-values.yaml")}"]
   create_namespace  = false
   recreate_pods     = true
   force_update      = true

terraform/via-provider-oke-ons-webhook.tf

@@ -6,7 +6,7 @@ resource "helm_release" "oke-ons-webhook" {
   depends_on        = [helm_release.prometheus]
   namespace         = var.monitoring_namespace
   name              = "oke-ons-webhook"
-  chart             = "${path.root}/files/oke-ons-webhook"
+  chart             = "${path.module}/files/oke-ons-webhook"
   version           = var.oke_ons_webhook_chart_version
   set              = [
     {