Add a way to invalidate docker images cache

docker/docs/builder/Dockerfile

@@ -1,15 +1,16 @@
 # rebuild in #36968
 # docker build -t clickhouse/docs-builder .
 # nodejs 17 prefers ipv6 and is broken in our environment
-FROM node:16.14.2-alpine3.15
+FROM node:16-alpine
 
 RUN apk add --no-cache git openssh bash
 
-# TODO: clean before merge!
-ARG DOCS_BRANCH=main
+# At this point we want to really update /opt/clickhouse-docs
+# despite the cached images
+ARG CACHE_INVALIDATOR=0
 
 RUN git clone https://github.com/ClickHouse/clickhouse-docs.git \
-    --depth=1 --branch=${DOCS_BRANCH} /opt/clickhouse-docs
+    --depth=1 --branch=main /opt/clickhouse-docs
 
 WORKDIR /opt/clickhouse-docs
 

docker/docs/builder/run.sh

@@ -8,8 +8,6 @@ if [ "$GIT_DOCS_BRANCH" ] && ! [ "$GIT_DOCS_BRANCH" == "$GIT_BRANCH" ]; then
   git fetch origin --depth=1 -- "$GIT_DOCS_BRANCH:$GIT_DOCS_BRANCH"
   git checkout "$GIT_DOCS_BRANCH"
 else
-  # Untracked yarn.lock could cause pull to fail
-  git clean -fdx
   # Update docs repo
   git pull
 fi

tests/ci/docker_images_check.py

@@ -12,14 +12,14 @@ from typing import Dict, List, Optional, Set, Tuple, Union
 
 from github import Github
 
-from env_helper import GITHUB_WORKSPACE, RUNNER_TEMP, GITHUB_RUN_URL
-from s3_helper import S3Helper
-from pr_info import PRInfo
-from get_robot_token import get_best_robot_token, get_parameter_from_ssm
-from upload_result_helper import upload_results
-from commit_status_helper import post_commit_status
 from clickhouse_helper import ClickHouseHelper, prepare_tests_results_for_clickhouse
+from commit_status_helper import post_commit_status
+from env_helper import GITHUB_WORKSPACE, RUNNER_TEMP, GITHUB_RUN_URL
+from get_robot_token import get_best_robot_token, get_parameter_from_ssm
+from pr_info import PRInfo
+from s3_helper import S3Helper
 from stopwatch import Stopwatch
+from upload_result_helper import upload_results
 
 NAME = "Push to Dockerhub (actions)"
 
@@ -237,6 +237,8 @@ def build_and_push_one_image(
             "docker buildx build --builder default "
             f"--label build-url={GITHUB_RUN_URL} "
             f"{from_tag_arg}"
+            # A hack to invalidate cache, grep for it in docker/ dir
+            f"--build-arg CACHE_INVALIDATOR={GITHUB_RUN_URL} "
             f"--tag {image.repo}:{version_string} "
             f"--cache-from type=registry,ref={image.repo}:{version_string} "
             f"--cache-from type=registry,ref={image.repo}:latest "

tests/ci/docker_test.py

@@ -124,6 +124,7 @@ class TestDockerImageCheck(unittest.TestCase):
         self.assertIn(
             f"docker buildx build --builder default --label build-url={GITHUB_RUN_URL} "
             "--build-arg FROM_TAG=version "
+            f"--build-arg CACHE_INVALIDATOR={GITHUB_RUN_URL} "
             "--tag name:version --cache-from type=registry,ref=name:version "
             "--cache-from type=registry,ref=name:latest "
             "--cache-to type=inline,mode=max --push --progress plain path",
@@ -142,6 +143,7 @@ class TestDockerImageCheck(unittest.TestCase):
         self.assertIn(
             f"docker buildx build --builder default --label build-url={GITHUB_RUN_URL} "
             "--build-arg FROM_TAG=version2 "
+            f"--build-arg CACHE_INVALIDATOR={GITHUB_RUN_URL} "
             "--tag name:version2 --cache-from type=registry,ref=name:version2 "
             "--cache-from type=registry,ref=name:latest "
             "--cache-to type=inline,mode=max --progress plain path",
@@ -159,6 +161,7 @@ class TestDockerImageCheck(unittest.TestCase):
         mock_machine.assert_not_called()
         self.assertIn(
             f"docker buildx build --builder default --label build-url={GITHUB_RUN_URL} "
+            f"--build-arg CACHE_INVALIDATOR={GITHUB_RUN_URL} "
             "--tag name:version2 --cache-from type=registry,ref=name:version2 "
             "--cache-from type=registry,ref=name:latest "
             "--cache-to type=inline,mode=max --progress plain path",