consider runner host dir

Signed-off-by: Ettore Di Giacinto <[email protected]>

.github/workflows/test-gpu.yml

@@ -71,7 +71,10 @@ jobs:
           if [ ! -e /run/systemd/system ]; then
             sudo mkdir /run/systemd/system
           fi
-          make BUILD_TYPE=cublas prepare-e2e run-e2e-image test-e2e
+          make \
+            TEST_DIR="/host/tests/${{ github.head_ref || github.ref }}" \
+            BUILD_TYPE=cublas \
+            prepare-e2e run-e2e-image test-e2e
       - name: Release space from worker ♻
         if: always()
         run: |
@@ -82,5 +85,7 @@ jobs:
           sudo cat logs.txt || true
           sudo rm -rf logs.txt
           make clean || true
-          make teardown-e2e || true
+          make \
+            TEST_DIR="/host/tests/${{ github.head_ref || github.ref }}" \
+            teardown-e2e || true
           docker system prune -f -a --volumes || true

Dockerfile

@@ -166,14 +166,14 @@ COPY --from=builder /build/local-ai ./
 # do not let stablediffusion rebuild (requires an older version of absl)
 COPY --from=builder /build/backend-assets/grpc/stablediffusion ./backend-assets/grpc/stablediffusion
 
-# To run only if we are building extras
 # Copy VALLE-X as it's not a real "lib"
-# and we also copy exllama libs over to resolve exllama import error
-RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+RUN if [ -d /usr/lib/vall-e-x ]; then \
     cp -rfv /usr/lib/vall-e-x/* ./ ; \ 
-    if [ "${BUILD_TYPE}" = "cublas" ] && [ "${TARGETARCH:-$(go env GOARCH)}" = "amd64" ]; then \
+    fi
+
+# we also copy exllama libs over to resolve exllama import error
+RUN if [ -d /usr/local/lib/python3.9/dist-packages/exllama ]; then \
         cp -rfv /usr/local/lib/python3.9/dist-packages/exllama extra/grpc/exllama/;\
-    fi ;\
     fi
 
 # Define the health check command

Makefile

@@ -47,6 +47,8 @@ CUDA_LIBPATH?=/usr/local/cuda/lib64/
 GO_TAGS?=
 BUILD_ID?=git
 
+TEST_DIR=/tmp/test
+
 RANDOM := $(shell bash -c 'echo $$RANDOM')
 
 VERSION?=$(shell git describe --always --tags || echo "dev" )
@@ -335,12 +337,14 @@ test: prepare test-models/testmodel grpcs
 	$(MAKE) test-stablediffusion
 
 prepare-e2e:
-	test -e $(abspath ./tests/e2e-fixtures)/ggllm-test-model.bin || wget -q https://huggingface.co/TheBloke/CodeLlama-7B-Instruct-GGUF/resolve/main/codellama-7b-instruct.Q2_K.gguf -O $(abspath ./tests/e2e-fixtures)/ggllm-test-model.bin
+	mkdir -p $(TEST_DIR)
+	cp -rfv $(abspath ./tests/e2e-fixtures)/gpu.yaml $(TEST_DIR)/gpu.yaml
+	test -e $(TEST_DIR)/ggllm-test-model.bin || wget -q https://huggingface.co/TheBloke/CodeLlama-7B-Instruct-GGUF/resolve/main/codellama-7b-instruct.Q2_K.gguf -O $(TEST_DIR)/ggllm-test-model.bin
 	docker build --build-arg BUILD_GRPC=true --build-arg GRPC_BACKENDS="$(GRPC_BACKENDS)" --build-arg IMAGE_TYPE=core --build-arg BUILD_TYPE=$(BUILD_TYPE) --build-arg CUDA_MAJOR_VERSION=11 --build-arg CUDA_MINOR_VERSION=7 --build-arg FFMPEG=true -t localai-tests .
 
 run-e2e-image:
 	ls -liah $(abspath ./tests/e2e-fixtures)
-	docker run -p 5390:8080 -e MODELS_PATH=/models -e THREADS=1 -e DEBUG=true -d --rm -v $(abspath ./tests/e2e-fixtures):/models --gpus all --name e2e-tests-$(RANDOM) localai-tests
+	docker run -p 5390:8080 -e MODELS_PATH=/models -e THREADS=1 -e DEBUG=true -d --rm -v $(TEST_DIR):/models --gpus all --name e2e-tests-$(RANDOM) localai-tests
 
 test-e2e:
 	@echo 'Running e2e tests'
@@ -349,7 +353,7 @@ test-e2e:
 	$(GOCMD) run github.com/onsi/ginkgo/v2/ginkgo --flake-attempts 5 -v -r ./tests/e2e
 
 teardown-e2e:
-	rm -rf ./tests/e2e-fixtures/ggllm-test-model.bin
+	rm -rf $(TEST_DIR) || true
 	docker stop $$(docker ps -q --filter ancestor=localai-tests)
 
 test-gpt4all: prepare-test