Slim down docker cache size (#3190)

* slim down docker cache size

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* remove old centos images

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* troubleshoot test failure

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* fix wget version ref

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* refactor caching mechanisms

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* add cache cleanup steps

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* simplify deleting cache

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* fix first clone issue

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

* add tool dep

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

---------

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>
This commit is contained in:
Alex Goodman
2024-09-09 11:15:13 -04:00
committed by GitHub
parent deabd4115a
commit 0a3f513f92
75 changed files with 1302 additions and 382 deletions
+222 -54
View File
@@ -4,9 +4,19 @@ vars:
OWNER: anchore
PROJECT: syft
CACHE_IMAGE: ghcr.io/{{ .OWNER }}/{{ .PROJECT }}/test-fixture-cache:latest
# static file dirs
TOOL_DIR: .tool
TMP_DIR: .tmp
ORAS_CACHE: "{{ .TMP_DIR }}/oras-cache"
CACHE_PATHS_FILE: "{{ .TMP_DIR }}/cache_paths.json"
LAST_CACHE_PULL_FILE: "{{ .TMP_DIR }}/last_cache_paths.json"
# TOOLS
ORAS: "{{ .TOOL_DIR }}/oras"
YQ: "{{ .TOOL_DIR }}/yq"
TASK: "{{ .TOOL_DIR }}/task"
# used for changelog generation
CHANGELOG: CHANGELOG.md
@@ -33,6 +43,9 @@ vars:
COMPARE_DIR: ./test/compare
COMPARE_TEST_IMAGE: centos:8.2.2004
env:
GNUMAKEFLAGS: '--no-print-directory'
tasks:
## High-level tasks #################################
@@ -65,6 +78,7 @@ tasks:
- task: benchmark
- task: test-utils
- task: cli
- task: check-docker-cache
## Bootstrap tasks #################################
@@ -212,10 +226,6 @@ tasks:
# that the cache being restored with the correct binary will be rebuilt since the timestamps
# and local checksums will not line up.
deps: [tools, snapshot]
sources:
- "{{ .SNAPSHOT_BIN }}"
- ./test/cli/**
- ./**/*.go
cmds:
- cmd: "echo 'testing binary: {{ .SNAPSHOT_BIN }}'"
silent: true
@@ -229,18 +239,14 @@ tasks:
test-utils:
desc: Run tests for pipeline utils
sources:
- .github/scripts/labeler*.py
cmds:
- cmd: python .github/scripts/labeler_test.py
- cmd: .github/scripts/labeler_test.py
## Benchmark test targets #################################
benchmark:
deps: [tmpdir]
sources:
- ./**/*.go
generates:
- "{{ .TMP_DIR }}/benchmark-main.txt"
cmds:
@@ -253,8 +259,6 @@ tasks:
show-benchstat:
deps: [benchmark, tmpdir]
sources:
- "{{ .TMP_DIR }}/benchstat.txt"
cmds:
- cmd: "cat {{ .TMP_DIR }}/benchstat.txt"
silent: true
@@ -263,56 +267,188 @@ tasks:
## Test-fixture-related targets #################################
fingerprints:
desc: Generate test fixture fingerprints
desc: Generate fingerprints for all non-docker test fixture
silent: true
# this will look for `test-fixtures/Makefile` and invoke the `fingerprint` target to calculate all cache input fingerprint files
generates:
- cmd/syft/internal/test/integration/test-fixtures/cache.fingerprint
- syft/file/cataloger/executable/test-fixtures/elf/cache.fingerprint
- syft/file/cataloger/executable/test-fixtures/shared-info/cache.fingerprint
- syft/pkg/cataloger/binary/test-fixtures/cache.fingerprint
- syft/pkg/cataloger/java/test-fixtures/java-builds/cache.fingerprint
- syft/pkg/cataloger/golang/test-fixtures/archs/binaries.fingerprint
- syft/pkg/cataloger/redhat/test-fixtures/rpms.fingerprint
- syft/pkg/cataloger/kernel/test-fixtures/cache.fingerprint
- '**/test-fixtures/**/*.fingerprint'
- test/install/cache.fingerprint
- test/cli/test-fixtures/cache.fingerprint
cmds:
# for EXECUTABLE unit test fixtures
- "cd syft/file/cataloger/executable/test-fixtures/elf && make cache.fingerprint"
- "cd syft/file/cataloger/executable/test-fixtures/shared-info && make cache.fingerprint"
# for IMAGE integration test fixtures
- "cd cmd/syft/internal/test/integration/test-fixtures && make cache.fingerprint"
# for BINARY unit test fixtures
- "cd syft/pkg/cataloger/binary/test-fixtures && make cache.fingerprint"
# for JAVA BUILD unit test fixtures
- "cd syft/pkg/cataloger/java/test-fixtures/java-builds && make cache.fingerprint"
# for GO BINARY unit test fixtures
- "cd syft/pkg/cataloger/golang/test-fixtures/archs && make binaries.fingerprint"
# for RPM unit test fixtures
- "cd syft/pkg/cataloger/redhat/test-fixtures && make rpms.fingerprint"
# for Kernel unit test fixtures
- "cd syft/pkg/cataloger/kernel/test-fixtures && make cache.fingerprint"
# for INSTALL test fixtures
- "cd test/install && make cache.fingerprint"
# for CLI test fixtures
- "cd test/cli/test-fixtures && make cache.fingerprint"
- |
BOLD='\033[1m'
YELLOW='\033[0;33m'
RESET='\033[0m'
fixtures:
desc: Generate test fixtures
echo -e "${YELLOW}creating fingerprint files for non-docker fixtures...${RESET}"
for dir in $(find . -type d -name 'test-fixtures'); do
if [ -f "$dir/Makefile" ]; then
# for debugging...
#echo -e "${YELLOW}• calculating fingerprints in $dir... ${RESET}"
(make -C "$dir" fingerprint)
fi
done
# for debugging...
# echo -e "generated all fixture fingerprints"
- .github/scripts/fingerprint_docker_fixtures.py
- |
# if DOWNLOAD_TEST_FIXTURE_CACHE is set to 'false', then we don't need to calculate the fingerprint for the cache
if [ "$DOWNLOAD_TEST_FIXTURE_CACHE" = "false" ]; then
exit 0
fi
.github/scripts/find_cache_paths.py {{ .CACHE_PATHS_FILE }} > /dev/null
refresh-fixtures:
desc: Clear and fetch all test fixture cache
aliases:
- fixtures
silent: true
deps:
- tools
cmds:
- "cd syft/file/cataloger/executable/test-fixtures/elf && make"
- "cd syft/file/cataloger/executable/test-fixtures/shared-info && make"
- "cd syft/pkg/cataloger/java/test-fixtures/java-builds && make"
- "cd syft/pkg/cataloger/redhat/test-fixtures && make"
- "cd syft/pkg/cataloger/binary/test-fixtures && make"
- |
BOLD='\033[1m'
PURPLE='\033[0;35m'
RESET='\033[0m'
# if DOWNLOAD_TEST_FIXTURE_CACHE is set to 'false', then skip the cache download and always build
if [ "$DOWNLOAD_TEST_FIXTURE_CACHE" = "false" ]; then
echo -e "${BOLD}${PURPLE}skipping cache download, rebuilding cache...${RESET}"
{{ .TASK }} build-fixtures
exit 0
fi
LATEST_FINGERPRINT=$(docker manifest inspect {{ .CACHE_IMAGE }} | {{ .YQ }} -r '.annotations.fingerprint')
echo "latest cache: $LATEST_FINGERPRINT"
if [ -f {{ .LAST_CACHE_PULL_FILE }} ]; then
LAST_PULL_FINGERPRINT=$(cat {{ .LAST_CACHE_PULL_FILE }} | {{ .YQ }} -r '.digest')
else
echo -e "${BOLD}${PURPLE}empty cache, downloading cache...${RESET}"
{{ .TASK }} download-test-fixture-cache
exit 0
fi
{{ .TASK }} fingerprints
WANT_FINGERPRINT=$(cat {{ .CACHE_PATHS_FILE }} | {{ .YQ }} -r '.digest')
echo "desired cache: $WANT_FINGERPRINT"
echo "last pulled cache: $LAST_PULL_FINGERPRINT"
# if we already have the latest cache, skip the refresh
if [ "$LAST_PULL_FINGERPRINT" = "$WANT_FINGERPRINT" ]; then
echo -e "${BOLD}${PURPLE}already have the latest cache (skipping cache download)${RESET}"
exit 0
fi
# at this point we only refresh the cache if we want the same cache that is currently available.
# we don't by default refresh the cache if the cache if it is simply different from what we have,
# because we may be working on a code change that doesn't require a cache refresh (but could trigger one,
# which would be annoying to deal with in a development workflow).
if [ "$LATEST_FINGERPRINT" = "$WANT_FINGERPRINT" ]; then
echo -e "${BOLD}${PURPLE}found newer cache! downloading cache...${RESET}"
{{ .TASK }} download-test-fixture-cache
else
echo -e "${BOLD}${PURPLE}found different cache, but isn't clear if it's newer (skipping cache download and manually building)${RESET}"
{{ .YQ }} eval '.paths[] | "\(.digest) \(.path)"' {{ .LAST_CACHE_PULL_FILE }} > .tmp/last_cache_lines
{{ .YQ }} eval '.paths[] | "\(.digest) \(.path)"' {{ .CACHE_PATHS_FILE }} > .tmp/cache_lines
diff .tmp/last_cache_lines .tmp/cache_lines || true
echo -e "${BOLD}${PURPLE}diff with more context...${RESET}"
diff -U10000 {{ .LAST_CACHE_PULL_FILE }} {{ .CACHE_PATHS_FILE }} || true
echo -e "${BOLD}${PURPLE}detected changes to input material, manually building fixtures...${RESET}"
{{ .TASK }} build-fixtures
fi
build-fixtures:
desc: Generate all non-docker test fixtures
silent: true
# this will look for `test-fixtures/Makefile` and invoke the `fixtures` target to generate any and all test fixtures
cmds:
- |
BOLD='\033[1m'
YELLOW='\033[0;33m'
RESET='\033[0m'
# Use a for loop with command substitution to avoid subshell issues
for dir in $(find . -type d -name 'test-fixtures'); do
if [ -f "$dir/Makefile" ]; then
echo -e "${YELLOW}${BOLD}generating fixtures in $dir${RESET}"
(make -C "$dir" fixtures)
fi
done
echo -e "${BOLD}generated all fixtures${RESET}"
download-test-fixture-cache:
desc: Download test fixture cache from ghcr.io
deps: [tools, clean-cache]
vars:
CACHE_DIGEST:
sh: docker manifest inspect {{ .CACHE_IMAGE }} | {{ .YQ }} -r '.annotations.fingerprint'
cmds:
- silent: true
cmd: |
# if oras cache is > 4 GB, delete it
if [ -d {{ .ORAS_CACHE }} ]; then
total_size=$(du -c {{ .ORAS_CACHE }} | grep total | awk '{print $1}')
if [ "$total_size" -gt 4194304 ]; then
echo 'deleting oras cache'
rm -rf {{ .ORAS_CACHE }}
fi
fi
- "ORAS_CACHE={{ .ORAS_CACHE }} {{ .ORAS }} pull {{ .CACHE_IMAGE }}"
- "cp {{ .CACHE_PATHS_FILE }} {{ .LAST_CACHE_PULL_FILE }}"
upload-test-fixture-cache:
desc: Upload the test fixture cache to ghcr.io
deps: [tools, fingerprints]
silent: true
cmd: |
set -eu
oras_command="{{ .ORAS }} push {{ .CACHE_IMAGE }}"
paths=$(cat {{ .CACHE_PATHS_FILE }} | {{ .YQ }} -r '.paths[].path')
for path in $paths; do
oras_command+=" $path"
done
oras_command+=" {{ .CACHE_PATHS_FILE }}"
oras_command+=" --annotation org.opencontainers.image.source=https://github.com/{{ .OWNER }}/{{ .PROJECT }}"
oras_command+=" --annotation fingerprint=$(cat {{ .CACHE_PATHS_FILE }} | {{ .YQ }} -r '.digest')"
echo "Executing: $oras_command"
eval $oras_command
show-test-image-cache:
silent: true
cmds:
- "echo '\nDocker daemon cache:'"
- "echo 'Docker daemon cache:'"
- "docker images --format '{{`{{.ID}}`}} {{`{{.Repository}}`}}:{{`{{.Tag}}`}}' | grep stereoscope-fixture- | sort"
- "echo '\nTar cache:'"
- 'find . -type f -wholename "**/test-fixtures/snapshot/*" | sort'
- 'find . -type f -wholename "**/test-fixtures/cache/stereoscope-fixture-*.tar" | sort'
check-docker-cache:
desc: Ensure docker caches aren't using too much disk space
silent: true
cmd: |
total_size=$(find . | grep cache | grep tar | xargs du -c | grep total | awk '{print $1}')
find . | grep cache | grep tar | xargs du
echo "total $total_size KB"
if [ "$total_size" -gt 1048576 ]; then
echo 'docker cache is larger than 1GB'
exit 1
fi
## install.sh testing targets #################################
@@ -457,7 +593,16 @@ tasks:
ci-check:
# desc: "[CI only] Are you in CI?"
cmds:
- cmd: .github/scripts/ci-check.sh
- cmd: |
red=$(tput setaf 1)
bold=$(tput bold)
normal=$(tput sgr0)
# assert we are running in CI (or die!)
if [[ -z "$CI" ]]; then
echo "${bold}${red}This step should ONLY be run in CI. Exiting...${normal}"
exit 1
fi
silent: true
ci-release:
@@ -489,8 +634,31 @@ tasks:
- "rm -rf {{ .SNAPSHOT_DIR }}"
- "rm -rf {{ .TMP_DIR }}/goreleaser.yaml"
clean-cache:
desc: Remove all docker cache and local image tar cache
clean-docker-cache:
desc: Remove all docker cache tars and images from the daemon
cmds:
- 'find . -type f -wholename "**/test-fixtures/cache/stereoscope-fixture-*.tar" -delete'
- "docker images --format '{{`{{.ID}}`}} {{`{{.Repository}}`}}' | grep stereoscope-fixture- | awk '{print $$1}' | uniq | xargs -r docker rmi --force"
- find . -type d -wholename "**/test-fixtures/cache" | xargs rm -rf
- docker images --format '{{`{{.ID}}`}} {{`{{.Repository}}`}}' | grep stereoscope-fixture- | awk '{print $1}' | uniq | xargs -r docker rmi --force
clean-oras-cache:
desc: Remove all cache for oras commands
cmd: rm -rf {{ .ORAS_CACHE }}
clean-cache:
desc: Remove all image docker tar cache, images from the docker daemon, and ephemeral test fixtures
cmds:
- task: clean-docker-cache
- |
BOLD='\033[1m'
YELLOW='\033[0;33m'
RESET='\033[0m'
# Use a for loop with command substitution to avoid subshell issues
for dir in $(find . -type d -name 'test-fixtures'); do
if [ -f "$dir/Makefile" ]; then
echo -e "${YELLOW}${BOLD}deleting ephemeral test fixtures in $dir${RESET}"
(make -C "$dir" clean)
fi
done
echo -e "${BOLD}Deleted all ephemeral test fixtures${RESET}"
- rm -f {{ .LAST_CACHE_PULL_FILE }} {{ .CACHE_PATHS_FILE }}