Consolidating images and artifacts

.github/actions/docker-build-and-push/action.yml

@@ -33,6 +33,10 @@ inputs:
     required: false
     description: Git commit SHA to embed via COMMIT_SHA build arg
     default: ''
+  additional-tags:
+    required: false
+    description: Additional image tags to push
+    default: ''
 runs:
   using: "composite"
   steps:
@@ -58,6 +62,7 @@ runs:
           ${{ inputs.registry }}/${{ inputs.image-name }}:${{ inputs.tag }}
           ${{ inputs.push == 'true' && inputs.prerelease != 'true' && format('{0}/{1}:latest', inputs.registry, inputs.image-name) || '' }}
           ${{ inputs.commit-sha != '' && format('{0}/{1}:{2}', inputs.registry, inputs.image-name, inputs.commit-sha) || '' }}
+          ${{ inputs.additional-tags }}
         build-args: |
           LDFLAGS=-s -w
           COMMIT_SHA=${{ inputs.commit-sha || 'unknown' }}

.github/actions/helm-build-and-push/action.yml

@@ -12,7 +12,7 @@ inputs:
     required: false
     default: ''
   epp_image_repository:
-    description: 'EPP image repository name (e.g. llm-d-router-endpoint-picker or llm-d-router-endpoint-picker-dev)'
+    description: 'EPP image repository name (e.g. llm-d-router-endpoint-picker)'
     required: false
     default: 'llm-d-router-endpoint-picker'
 runs:

.github/workflows/ci-build-images.yaml

@@ -19,6 +19,14 @@ on:
         required: false
         type: string
         default: ''
+      epp-additional-tags:
+        required: false
+        type: string
+        default: ''
+      sidecar-additional-tags:
+        required: false
+        type: string
+        default: ''
 
 permissions:
   contents: read
@@ -60,6 +68,7 @@ jobs:
           prerelease: ${{ inputs.prerelease }}
           commit-sha: ${{ github.sha }}
           push: 'true'
+          additional-tags: ${{ inputs.epp-additional-tags }}
 
   build-sidecar:
     runs-on: ubuntu-latest
@@ -93,6 +102,7 @@ jobs:
           prerelease: ${{ inputs.prerelease }}
           commit-sha: ${{ github.sha }}
           push: 'true'
+          additional-tags: ${{ inputs.sidecar-additional-tags }}
 
   push-helm-charts:
     needs: [build-epp, build-sidecar]

.github/workflows/ci-dev.yaml

@@ -19,6 +19,8 @@ jobs:
     outputs:
       epp_name: ${{ steps.version.outputs.epp_name }}
       sidecar_name: ${{ steps.version.outputs.sidecar_name }}
+      epp_base_name: ${{ steps.version.outputs.epp_base_name }}
+      sidecar_base_name: ${{ steps.version.outputs.sidecar_base_name }}
       tag: ${{ steps.tag.outputs.tag }}
     steps:
       - name: Set image names
@@ -27,6 +29,8 @@ jobs:
           repo="${GITHUB_REPOSITORY##*/}"
           echo "epp_name=${repo}-endpoint-picker-dev" >> "$GITHUB_OUTPUT"
           echo "sidecar_name=${repo}-disagg-sidecar-dev" >> "$GITHUB_OUTPUT"
+          echo "epp_base_name=${repo}-endpoint-picker" >> "$GITHUB_OUTPUT"
+          echo "sidecar_base_name=${repo}-disagg-sidecar" >> "$GITHUB_OUTPUT"
 
       - name: Set branch name as tag
         id: tag
@@ -42,5 +46,7 @@ jobs:
       sidecar-image-name: ${{ needs.set-params.outputs.sidecar_name }}
       tag: ${{ needs.set-params.outputs.tag }}
       prerelease: true
-      chart-suffix: "-dev"
+      chart-suffix: ""
+      epp-additional-tags: ghcr.io/llm-d/${{ needs.set-params.outputs.epp_base_name }}:main
+      sidecar-additional-tags: ghcr.io/llm-d/${{ needs.set-params.outputs.sidecar_base_name }}:main
 

config/charts/README.md

@@ -119,7 +119,7 @@ Core settings for the Endpoint Picker Proxy (EPP) container and pod, including s
 | `router.epp.replicas` | Number of EPP replicas. Set > 1 to enable multi-replica EPP. | `1` |
 | `router.epp.extProcPort` | Port EPP uses for external processing gRPC communication. | `9002` |
 | `router.epp.image.registry` | EPP container image registry. | `ghcr.io/llm-d` |
-| `router.epp.image.repository` | EPP container image repository. | `llm-d-router-endpoint-picker-dev` |
+| `router.epp.image.repository` | EPP container image repository. | `llm-d-router-endpoint-picker` |
 | `router.epp.image.tag` | EPP container image tag. | `main` |
 | `router.epp.image.pullPolicy` | EPP container image pull policy. | `Always` |
 | `router.epp.env` | Extra environment variables for EPP container. | `[]` |

config/charts/routerlib/values.yaml

@@ -7,7 +7,7 @@ epp:
   replicas: 1
   image:
     registry: ghcr.io/llm-d
-    repository: llm-d-router-endpoint-picker-dev
+    repository: llm-d-router-endpoint-picker
     tag: main
     pullPolicy: Always
   extProcPort: 9002

test/perf/config/router-configs/optimized-baseline.yaml

@@ -11,7 +11,7 @@ router:
     replicas: 1
     image:
       registry: ghcr.io/llm-d
-      repository: llm-d-router-endpoint-picker-dev
+      repository: llm-d-router-endpoint-picker
       tag: main
       pullPolicy: Always
     flags:

test/perf/run_nightly_perf.py

@@ -118,7 +118,7 @@ def deploy_epp(ns, chart_path, chart_version, router_config_path, epp_cpu="2", e
     release_name = os.path.splitext(os.path.basename(router_config_path))[0]
 
     epp_registry = os.environ.get("EPP_REGISTRY", "ghcr.io/llm-d")
-    epp_repository = os.environ.get("EPP_REPOSITORY", "llm-d-router-endpoint-picker-dev")
+    epp_repository = os.environ.get("EPP_REPOSITORY", "llm-d-router-endpoint-picker")
     epp_tag = os.environ.get("EPP_TAG", "main")
 
     cpu_limit = double_cpu(epp_cpu)
@@ -657,7 +657,7 @@ def main():
     # Resolve relative defaults
     default_sim_deploy = os.path.join(script_dir, "config", "llm-d-sim-deployment.yaml")
     default_sim_svc = os.path.join(script_dir, "config", "llm-d-sim-service.yaml")
-    default_router_chart = "oci://ghcr.io/llm-d/charts/llm-d-router-standalone-dev"
+    default_router_chart = "oci://ghcr.io/llm-d/charts/llm-d-router-standalone"
     default_perf_chart = os.path.abspath(os.path.join(script_dir, "..", "..", "..", "..", "inference-perf", "deploy", "inference-perf"))
     default_perf_job = os.path.join(script_dir, "config", "shared_prefix_job1.yaml")
     default_results_dir = os.path.abspath(os.path.join(script_dir, "results"))