ci: fix k8s fallback builder bootstrap (#8346)

Signed-off-by: rrubin <rrubin@nvidia.com>
Signed-off-by: Ran Rubin <ranrubin@gmail.com>

.github/actions/bootstrap-buildkit/action.yml

@@ -59,8 +59,8 @@ inputs:
     description: 'Skip the bootstrap step (only create the builder)'
     required: false
     default: 'false'
-  fresh_builder:
-    description: 'Force creation of a new K8s builder even if one already exists. Used by the create-fresh-builder preliminary job.'
+  suppress_fallback_warning:
+    description: 'Suppress the fallback pod warning when intentionally using the Kubernetes driver'
     required: false
     default: 'false'
 
@@ -87,13 +87,7 @@ runs:
       if: inputs.buildkit_worker_addresses == ''
       shell: bash
       run: |
-        if [[ "${{ inputs.fresh_builder }}" != "true" ]] && docker buildx inspect ${{ inputs.builder_name }} > /dev/null 2>&1; then
-            echo "✅ Builder '${{ inputs.builder_name }}' already exists. Skipping creation."
-        else
-          if [[ "${{ inputs.fresh_builder }}" == "true" ]] && docker buildx inspect ${{ inputs.builder_name }} > /dev/null 2>&1; then
-            echo "🔄 Forcing fresh K8s builder: removing existing '${{ inputs.builder_name }}'."
-            docker buildx rm ${{ inputs.builder_name }} || true
-          fi
+        # If the builder already exists, these commands will just set the right configurations to use it
         echo "🔨 Creating K8s builder '${{ inputs.builder_name }}'."
         docker buildx create --use --name ${{ inputs.builder_name }} --driver kubernetes --platform=linux/amd64 \
         '--driver-opt=requests.ephemeral-storage=${{ inputs.ephemeral_storage }}' \
@@ -116,17 +110,18 @@ runs:
         '--driver-opt=limits.memory=${{ inputs.limits_memory }}' \
         '--driver-opt="nodeselector=kubernetes.io/arch=arm64,role=dynamo-builder-fallback"' \
         '--driver-opt="tolerations=${{ inputs.tolerations }}"'
-        fi
+
         sleep 3 # Give the builders some time to be ready
 
-        if [[ "${{ inputs.skip_bootstrap }}" != "true" && "${{ inputs.fresh_builder }}" != "true" ]]; then
-          echo "::warning::Build is using fallback pod. Please alert the ops team."
+        if [[ "${{ inputs.suppress_fallback_warning }}" != "true" ]]; then
+          echo "::warning::Build is using fallback pod"
           echo "## ⚠️ Fallback Build Warning" >> $GITHUB_STEP_SUMMARY
-          echo "This build is running on a **fallback pod**. Please alert the ops team." >> $GITHUB_STEP_SUMMARY
+          echo "This build is using the **fallback pod** because the preferred remote builders are unavailable. Expected during nightly runs and outside business hours." >> $GITHUB_STEP_SUMMARY
           echo "" >> $GITHUB_STEP_SUMMARY
         fi
 
 
+
     - name: Bootstrap buildkit
       if: inputs.skip_bootstrap != 'true'
       shell: bash

.github/actions/init-dynamo-builder/action.yml

@@ -143,8 +143,7 @@ runs:
       uses: ./.github/actions/bootstrap-buildkit
       with:
         builder_name: ${{ inputs.builder_name }}
-        buildkit_worker_addresses: ${{ steps.prepare.outputs.worker_addresses }}
-        fresh_builder: ${{ inputs.fresh_builder }}
+        buildkit_worker_addresses: ${{ inputs.fresh_builder != 'true' && steps.prepare.outputs.worker_addresses || '' }}
         ephemeral_storage: ${{ inputs.ephemeral_storage }}
         namespace: ${{ inputs.namespace }}
         replicas: ${{ inputs.replicas }}
@@ -152,4 +151,4 @@ runs:
         requests_memory: ${{ inputs.requests_memory }}
         limits_memory: ${{ inputs.limits_memory }}
         tolerations: ${{ inputs.tolerations }}
-
+        suppress_fallback_warning: ${{ inputs.fresh_builder }}
\ No newline at end of file

.github/workflows/build-frontend-image.yaml

@@ -102,7 +102,7 @@ jobs:
       - name: Build EPP Image
         id: build-epp-image
         shell: bash
-        timeout-minutes: 20
+        timeout-minutes: 30
         run: |
           set -x
           EPP_REPOSITORY="${{ env.IMAGE_REGISTRY }}/${{ env.IMAGE_REPOSITORY }}/dynamo-epp"

.github/workflows/nightly-ci.yml

@@ -39,8 +39,7 @@ jobs:
         with:
           builder_name: ${{ steps.export-builder-name.outputs.builder_name }}
           buildkit_worker_addresses: ''
-          fresh_builder: 'true'
-
+          suppress_fallback_warning: 'true'
 
   # ============================================================================
   # FRAMEWORK PIPELINES (Build → Test → Copy)