Merge remote-tracking branch 'upstream/main' into rhoai-3.2

Author: m-rafeeq

.github/actions/setup-postgres/action.yml

@@ -0,0 +1,33 @@
+name: Setup PostgreSQL
+description: Start PostgreSQL for llama-stack
+runs:
+  using: "composite"
+  steps:
+    - name: Start PostgreSQL
+      shell: bash
+      run: |
+        docker run -d \
+          --name postgres \
+          --net=host \
+          -e POSTGRES_USER=llamastack \
+          -e POSTGRES_PASSWORD=llamastack \
+          -e POSTGRES_DB=llamastack \
+          postgres:17
+
+        # Wait for PostgreSQL to be ready
+        echo "Waiting for PostgreSQL to be ready..."
+        postgres_ready=false
+        for i in {1..30}; do
+          if docker exec postgres pg_isready -U llamastack -d llamastack; then
+            echo "PostgreSQL is ready!"
+            postgres_ready=true
+            break
+          fi
+          echo "Attempt $i: PostgreSQL not ready yet..."
+          sleep 1
+        done
+        if [ "$postgres_ready" != true ]; then
+          echo "PostgreSQL failed to start after 30s :("
+          docker logs postgres || true
+          exit 1
+        fi

.github/workflows/redhat-distro-container.yml

@@ -102,6 +102,11 @@ jobs:
         id: vllm
         uses: ./.github/actions/setup-vllm
 
+      - name: Setup PostgreSQL for llama-stack
+        if: github.event_name != 'workflow_dispatch'
+        id: postgres
+        uses: ./.github/actions/setup-postgres
+
       - name: Start and smoke test LLS distro image
         if: github.event_name != 'workflow_dispatch'
         id: smoke-test
@@ -123,6 +128,7 @@ jobs:
 
           docker logs llama-stack > logs/llama-stack.log 2>&1 || echo "Failed to get llama-stack logs" > logs/llama-stack.log
           docker logs vllm > logs/vllm.log 2>&1 || echo "Failed to get vllm logs" > logs/vllm.log
+          docker logs postgres > logs/postgres.log 2>&1 || echo "Failed to get postgres logs" > logs/postgres.log
 
           # Gather system information
           echo "=== System information ==="
@@ -157,7 +163,7 @@ jobs:
         if: always()
         shell: bash
         run: |
-          docker rm -f vllm llama-stack
+          docker rm -f vllm llama-stack postgres
 
       - name: Log in to Quay.io
         id: login

distribution/run.yaml

@@ -64,13 +64,13 @@ providers:
     config:
       db_path: /opt/app-root/src/.llama/distributions/rh/milvus.db
       persistence:
-        backend: kv_milvus_inline
+        backend: kv_default
         namespace: vector_io::milvus
   - provider_id: ${env.ENABLE_FAISS:+faiss}
     provider_type: inline::faiss
     config:
       persistence:
-        backend: kv_faiss
+        backend: kv_default
         namespace: vector_io::faiss
   - provider_id: ${env.MILVUS_ENDPOINT:+milvus-remote}
     provider_type: remote::milvus
@@ -83,7 +83,7 @@ providers:
       client_pem_path: ${env.MILVUS_CLIENT_PEM_PATH:=}
       client_key_path: ${env.MILVUS_CLIENT_KEY_PATH:=}
       persistence:
-        backend: kv_milvus_remote
+        backend: kv_default
         namespace: vector_io::milvus_remote
   - provider_id: ${env.ENABLE_PGVECTOR:+pgvector}
     provider_type: remote::pgvector
@@ -94,7 +94,7 @@ providers:
       user: ${env.PGVECTOR_USER:=}
       password: ${env.PGVECTOR_PASSWORD:=}
       persistence:
-        backend: kv_pgvector
+        backend: kv_default
         namespace: vector_io::pgvector
   safety:
     - provider_id: trustyai_fms
@@ -110,10 +110,10 @@ providers:
     config:
       persistence:
         agent_state:
-          backend: kv_agents
+          backend: kv_default
           namespace: agents::meta_reference
         responses:
-          backend: sql_agents
+          backend: sql_default
           table_name: agents_responses
           max_write_queue_size: 10000
           num_writers: 4
@@ -147,13 +147,13 @@ providers:
     provider_type: remote::huggingface
     config:
       kvstore:
-        backend: kv_datasetio_huggingface
+        backend: kv_default
         namespace: datasetio::huggingface
   - provider_id: localfs
     provider_type: inline::localfs
     config:
       kvstore:
-        backend: kv_datasetio_localfs
+        backend: kv_default
         namespace: datasetio::localfs
   scoring:
   - provider_id: basic
@@ -189,7 +189,7 @@ providers:
     config:
       storage_dir: /opt/app-root/src/.llama/distributions/rh/files
       metadata_store:
-        backend: sql_files
+        backend: sql_default
         table_name: files_metadata
   - provider_id: ${env.ENABLE_S3:+s3}
     provider_type: remote::s3
@@ -201,7 +201,7 @@ providers:
       endpoint_url: ${env.S3_ENDPOINT_URL:=}
       auto_create_bucket: ${env.S3_AUTO_CREATE_BUCKET:=false}
       metadata_store:
-        backend: sql_files
+        backend: sql_default
         table_name: files_metadata
   batches:
   - provider_id: reference
@@ -213,48 +213,27 @@ providers:
 storage:
   backends:
     kv_default:
-      type: kv_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/kvstore.db
-    kv_agents:
-      type: kv_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/agents_store.db
-    kv_faiss:
-      type: kv_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/faiss.db
-    kv_milvus_inline:
-      type: kv_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/milvus_inline_registry.db
-    kv_milvus_remote:
-      type: kv_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/milvus_remote_registry.db
-    kv_pgvector:
-      type: kv_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/pgvector_registry.db
-    kv_datasetio_huggingface:
-      type: kv_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/huggingface_datasetio.db
-    kv_datasetio_localfs:
-      type: kv_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/localfs_datasetio.db
-    sql_inference:
-      type: sql_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/inference_store.db
-    sql_agents:
-      type: sql_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/responses_store.db
-    sql_files:
-      type: sql_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/files_metadata.db
+      type: kv_postgres
+      host: ${env.POSTGRES_HOST:=localhost}
+      port: ${env.POSTGRES_PORT:=5432}
+      db: ${env.POSTGRES_DB:=llamastack}
+      user: ${env.POSTGRES_USER:=llamastack}
+      password: ${env.POSTGRES_PASSWORD:=llamastack}
+      table_name: ${env.POSTGRES_TABLE_NAME:=llamastack_kvstore}
     sql_default:
-      type: sql_sqlite
-      db_path: /opt/app-root/src/.llama/distributions/rh/sql_store.db
+      type: sql_postgres
+      host: ${env.POSTGRES_HOST:=localhost}
+      port: ${env.POSTGRES_PORT:=5432}
+      db: ${env.POSTGRES_DB:=llamastack}
+      user: ${env.POSTGRES_USER:=llamastack}
+      password: ${env.POSTGRES_PASSWORD:=llamastack}
   stores:
     metadata:
       backend: kv_default
       namespace: registry
     inference:
-      backend: sql_inference
       table_name: inference_store
+      backend: sql_default
       max_write_queue_size: 10000
       num_writers: 4
     conversations:

tests/smoke.sh

@@ -15,6 +15,11 @@ function start_and_wait_for_llama_stack_container {
     --env ENABLE_SENTENCE_TRANSFORMERS=True \
     --env EMBEDDING_PROVIDER=sentence-transformers \
     --env TRUSTYAI_LMEVAL_USE_K8S=False \
+    --env POSTGRES_HOST="${POSTGRES_HOST:-localhost}" \
+    --env POSTGRES_PORT="${POSTGRES_PORT:-5432}" \
+    --env POSTGRES_DB="${POSTGRES_DB:-llamastack}" \
+    --env POSTGRES_USER="${POSTGRES_USER:-llamastack}" \
+    --env POSTGRES_PASSWORD="${POSTGRES_PASSWORD:-llamastack}" \
     --name llama-stack \
     "$IMAGE_NAME:$GITHUB_SHA"
   echo "Started Llama Stack container..."
@@ -70,6 +75,65 @@ function test_model_openai_inference {
   fi
 }
 
+function test_postgres_tables_exist {
+  echo "===> Verifying PostgreSQL tables have been created..."
+
+  # Expected tables created by llama-stack
+  expected_tables=("llamastack_kvstore" "inference_store")
+
+  # Retry for up to 10 seconds for tables to be created
+  for i in {1..10}; do
+    tables=$(docker exec postgres psql -U llamastack -d llamastack -t -c "SELECT tablename FROM pg_tables WHERE schemaname = 'public';" 2>/dev/null | tr -d ' ' | tr '\n' ' ')
+    all_found=true
+    for table in "${expected_tables[@]}"; do
+      if ! echo "$tables" | grep -q "$table"; then
+        all_found=false
+        break
+      fi
+    done
+    if [ "$all_found" = true ]; then
+      echo "===> All expected tables found: ${expected_tables[*]}"
+      echo "===> Available tables: $tables"
+      return 0
+    fi
+    echo "Attempt $i: Waiting for tables to be created..."
+    sleep 1
+  done
+
+  echo "===> PostgreSQL tables not created after 10s :("
+  echo "Expected tables: ${expected_tables[*]}"
+  echo "Available tables: $tables"
+  docker exec postgres psql -U llamastack -d llamastack -c "\dt" || true
+  return 1
+}
+
+function test_postgres_populated {
+  echo "===> Verifying PostgreSQL database has been populated..."
+
+  # Check that chat_completions table has data (retry for up to 10 seconds)
+  echo "Waiting for inference_store table to be populated..."
+  for i in {1..10}; do
+    inference_count=$(docker exec postgres psql -U llamastack -d llamastack -t -c "SELECT COUNT(*) FROM inference_store;" 2>/dev/null | tr -d ' ')
+    if [ -n "$inference_count" ] && [ "$inference_count" -gt 0 ]; then
+      echo "===> inference_store table has $inference_count record(s)"
+      break
+    fi
+    echo "Attempt $i: inference_store table not yet populated..."
+    sleep 1
+  done
+  if [ -z "$inference_count" ] || [ "$inference_count" -eq 0 ]; then
+    echo "===> PostgreSQL inference_store table is empty or doesn't exist after 10s :("
+    echo "Tables in database:"
+    docker exec postgres psql -U llamastack -d llamastack -c "\dt" || true
+    echo "inference_store table contents:"
+    docker exec postgres psql -U llamastack -d llamastack -t -c "SELECT COUNT(*) FROM inference_store;" || true
+    return 1
+  fi
+
+  echo "===> PostgreSQL database verification passed :)"
+  return 0
+}
+
 main() {
   echo "===> Starting smoke test..."
   start_and_wait_for_llama_stack_container
@@ -78,6 +142,14 @@ main() {
     exit 1
   fi
   test_model_openai_inference
+  if ! test_postgres_tables_exist; then
+    echo "PostgreSQL tables verification failed :("
+    exit 1
+  fi
+  if ! test_postgres_populated; then
+    echo "PostgreSQL data verification failed :("
+    exit 1
+  fi
   echo "===> Smoke test completed successfully!"
 }