jeirmeister/Cockatrice
1
0
Fork 0
mirror of https://github.com/Cockatrice/Cockatrice.git synced 2026-11-07 18:43:26 -08:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

[CI] Cache the macOS legs again (#7805)

Browse source 
Re-enable ccache on the three pull-request macOS legs (14, 15 and 27
Debug), which #7792 dropped because none of them persisted a cache.
Every leg now restores and saves its own entry on master, keyed by
target, and the eviction step keeps each entry inside its bound.

The bounds are sized from the measured working set: one macOS
configuration wrote a 0.5 GB ccache directory at -g1 with the old 700M
bound, and the debug-information change builds the vcpkg legs at level
0, which shrinks it to ~0.4 GB. 700M holds the whole configuration for
each leg, and the three entries add ~1.2 GB.

That fits the quota once master's slimmed Qt archive set and the CodeQL
prune are counted: the live total before the compiler caches is 7.1 GB,
Windows adds 0.95 GB, and the three macOS entries add 1.23 GB, for
~9.3 GB of 10.
This commit is contained in:
BruebachL 2026-11-07 20:58:29 +01:00 • committed by GitHub
parent 1df8c8efcb
commit bf2529d768
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 53 additions and 31 deletions
Download patch file Download diff file Expand all files Collapse all files

84
.github/workflows/desktop-build.yml vendored
View file

@ -234,7 +234,7 @@ jobs:
# That is 10.30 GB against the 10 GB quota, so every save past this point costs some
# other entry, and least-recently-used order reaches the buildkit layers before the
# trap caches. prune-caches below keeps only the newest trap cache; the live total is
# then 7.1 GB, which is what leaves room for a Windows compiler cache.
# then 7.1 GB, which is what leaves room for the Windows and macOS compiler caches.
#
# Measured 2026-11-04: one configuration costs 1.2 GB with full DWARF and 660 MB with
# -g1 (COCKATRICE_DEBUG_INFO_LEVEL below), and the size of the cache decides the hit
@ -249,22 +249,25 @@ jobs:
# Fedora 45 debug Debug + suite 660 MB 850M
# Fedora 45 release Release + package 614 MB 800M
# Windows 10 Release + package 890 MB working set 950M
# macOS 14/15 Release + package 0.5 GB ccache at -g1 700M
# macOS 27 Debug Debug + suite 0.5 GB ccache at -g1 700M
#
# That is 4.25 GB of bounds; what lands in the quota is 3.9 GB of compiler cache,
# 2.97 GB stored by the Linux legs and 0.95 GB by Windows, for 8.1 GB of 10 in total.
# That is 6.35 GB of bounds; what lands in the quota is 5.15 GB of compiler cache,
# 2.97 GB stored by the Linux legs, 0.95 GB by Windows and 1.23 GB by the three macOS
# legs, for 9.3 GB of 10 in total.
# So only legs worth their share get one. A leg keeps no cache when it duplicates
# another leg's coverage (Ubuntu 26.04 behind Debian 13's Release + suite; the macOS
# legs behind the Linux ones) or finishes well inside the Windows leg (Servatrice
# covers the server alone in 4.7 min). A macOS
# cache served 42% and saved about 32s on a 20 min leg that is already shorter than
# Windows, so that bound is better spent here. The four Linux legs that stay compile
# the client on the toolchains a change can break and are the ones a cache shortens
# by the most.
# another leg's coverage (Ubuntu 26.04 behind Debian 13's Release + suite) or finishes
# well inside the Windows leg (Servatrice covers the server alone in 4.7 min). The
# macOS legs compile the client with a different toolchain (Apple clang), and one
# configuration writes only ~0.5 GB, so all three fit beside the Windows entry even
# though a macOS cache served 42% and saved about 32s on a 20 min leg. The four Linux
# legs that stay compile the client on the toolchains a change can break and are the
# ones a cache shortens by the most.
# Measured job totals on master 2026-11-04 also show the ceiling this leaves:
# Windows 10 is 29.5 min and compiles for 23.7 of it, so Windows sets the
# pull-request wall clock no matter how warm these legs are. That is why the Windows
# leg is the one macOS's share went to (see build-vcpkg): MSVC has no -g level to
# shrink, and with /Z7 the whole configuration is cacheable, PCH included.
# leg keeps the largest bound (see build-vcpkg): MSVC has no -g level to shrink, and
# with /Z7 the whole configuration is cacheable, PCH included.
#
# Steady-state master data corrected this on 2026-11-07: Arch 90.1% and Debian 13
# 96.75% per-run hits at the 850M/800M bounds, Fedora 45 debug 99.18% and release
@ -273,8 +276,9 @@ jobs:
# run on the same day measured one configuration at 680 MB (Arch) and 625 MB
# (Debian 13), so those bounds already hold the working set with ~170 MB of slack for
# commit-to-commit churn, and raising them buys hit rate the numbers do not show a
# need for. macOS 14's 800M entry is dropped because persistence buys 32s there; the
# quota stays free for the legs whose bounds would actually miss.
# need for. The three macOS entries are 700M each because one configuration wrote a
# 0.5 GB ccache directory at -g1 and level 0 shrinks it further, so the bound holds the
# whole configuration with room to spare.
#
# Re-add a leg only with a bound that holds its whole configuration: a smaller one is
# worse than none.
@ -625,6 +629,8 @@ jobs:
target: 14 # EOL 2026-09-14
runner: xcode-27 # https://github.com/actions/runner-images/blob/main/images/macos/xcode-27-arm64-Readme.md
ccache_eviction_age: 7d
ccache_size: 700M # Release + package: ~0.4 GB at level 0
cmake_generator: Ninja
make_package: 1
override_target: 14
@ -633,6 +639,7 @@ jobs:
qt_modules: qtimageformats qtmultimedia qtshadertools qtwebsockets
soc: Apple
type: Release
use_ccache: 1
vcpkg_deployment_target: 14
vcpkg_scope: 'vcpkg/macos14-arm64-xcode26.6/'
vcpkg_upload: 1 # only one leg per vcpkg_scope hands its archives over
@ -641,6 +648,8 @@ jobs:
target: 15 # EOL 2027-??
runner: xcode-27 # https://github.com/actions/runner-images/blob/main/images/macos/xcode-27-arm64-Readme.md
ccache_eviction_age: 7d
ccache_size: 700M # Release + package: ~0.4 GB at level 0
cmake_generator: Ninja
make_package: 1
override_target: 15
@ -649,6 +658,7 @@ jobs:
qt_modules: qtimageformats qtmultimedia qtshadertools qtwebsockets
soc: Apple
type: Release
use_ccache: 1
vcpkg_deployment_target: 14
vcpkg_scope: 'vcpkg/macos14-arm64-xcode26.6/'
@ -656,11 +666,14 @@ jobs:
target: 27
runner: xcode-27 # https://github.com/actions/runner-images/blob/main/images/macos/xcode-27-arm64-Readme.md
ccache_eviction_age: 7d
ccache_size: 700M # Debug + suite: ~0.5 GB at level 0
cmake_generator: Ninja
qt_version: 6.11.*
qt_modules: qtimageformats qtmultimedia qtshadertools qtwebsockets
soc: Apple
type: Debug
use_ccache: 1
vcpkg_deployment_target: 14
vcpkg_scope: 'vcpkg/macos14-arm64-xcode26.6/'
@ -695,10 +708,13 @@ jobs:
runs-on: ${{ matrix.runner }}
timeout-minutes: 100
env:
# Only the Windows leg keeps a compiler cache; the macOS legs build without one
# (see the note where their ccache steps used to be). ccache drives MSVC as long as
# the compiler emits /Z7, which CMakeLists.txt sets for exactly this reason, and it
# caches the PCH-backed targets with the sloppiness that .ci/compile.sh re-asserts.
# The Windows leg and all three pull-request macOS legs keep a compiler cache. ccache
# drives MSVC as long as the compiler emits /Z7, which CMakeLists.txt sets for exactly
# this reason, and it drives clang on macOS the same way; both cache the PCH-backed
# targets with the sloppiness that .ci/compile.sh re-asserts. Every leg builds without
# debug information on non-tag refs, so one macOS configuration writes ~0.4-0.5 GB and
# the 700M bound holds it with room for commit-to-commit churn (see build-linux for the
# quota arithmetic).
CCACHE_DIR: ${{ github.workspace }}/.cache/
CCACHE_SIZE: ${{ matrix.ccache_size || '700M' }}
PERSIST_CCACHE: ${{ matrix.persist_ccache || 'true' }}
@ -726,19 +742,25 @@ jobs:
shell: bash
run: choco install ccache --no-progress -y
- name: "[Windows] Restore compiler cache (ccache)"
if: matrix.os == 'Windows' && matrix.use_ccache == 1 && env.PERSIST_CCACHE == 'true'
- name: "[macOS] Setup ccache"
if: matrix.os == 'macOS' && matrix.use_ccache == 1
run: brew install ccache
- name: "Restore compiler cache (ccache)"
if: matrix.use_ccache == 1 && env.PERSIST_CCACHE == 'true'
id: restore_ccache
uses: actions/cache/restore@v6
with:
key: ccache-${{ matrix.os }}${{ matrix.target }}
path: ${{ env.CCACHE_DIR }}
# No macOS leg runs ccache. A restored macOS cache served 42% of calls and saved
# about 32s on a 20 min leg that is already shorter than the Windows leg, which did
# not pay for its 800M slice of the quota; with nothing restored, the remaining
# intra-run hits are 3% and the hashing is pure overhead on ~1160 compiles, so the
# steps that installed and restored ccache are gone rather than merely disabled.
# Every macOS leg that keeps a cache restores the master entry for its own target.
# Measured on master 2026-11-07 (before the debug-information change), one macOS
# configuration wrote a 0.5 GB ccache directory at -g1 against the 700M bound, and the
# restored cache served 42% of calls. Level 0 shrinks that directory to ~0.4 GB, so the
# bound holds the whole configuration and the three entries fit the quota (see the
# sizing note in build-linux). macOS 14 and 15 build the same Release + package
# configuration for two deployment targets, but each is cheap enough to keep its own.
- name: "[macOS] Install aqtinstall"
if: matrix.os == 'macOS'
@ -1079,10 +1101,10 @@ jobs:
retention-days: 1
# Delete used cache to emulate a ccache update. See https://github.com/actions/cache/issues/342
- name: "[Windows] Delete remote compiler cache (ccache)"
if: matrix.os == 'Windows' && matrix.use_ccache == 1 && env.PERSIST_CCACHE == 'true' && github.ref == 'refs/heads/master' && steps.restore_ccache.outputs.cache-hit
- name: "Delete remote compiler cache (ccache)"
if: matrix.use_ccache == 1 && env.PERSIST_CCACHE == 'true' && github.ref == 'refs/heads/master' && steps.restore_ccache.outputs.cache-hit
continue-on-error: true
# Bash, not the runner's default pwsh, which cannot parse this `if`.
# Bash, not the Windows runner's default pwsh, which cannot parse this `if`.
shell: bash
env:
CACHE_PRIMARY_KEY: ${{ steps.restore_ccache.outputs.cache-primary-key }}
@ -1092,8 +1114,8 @@ jobs:
echo "Cache deleted successfully"
fi
- name: "[Windows] Cache updated compiler cache (ccache)"
if: matrix.os == 'Windows' && matrix.use_ccache == 1 && env.PERSIST_CCACHE == 'true' && github.ref == 'refs/heads/master'
- name: "Cache updated compiler cache (ccache)"
if: matrix.use_ccache == 1 && env.PERSIST_CCACHE == 'true' && github.ref == 'refs/heads/master'
uses: actions/cache/save@v6
with:
key: ${{ steps.restore_ccache.outputs.cache-primary-key }}
@ -1287,7 +1309,7 @@ jobs:
# The nightly legs and the pull-request legs that finish under the Windows leg
# keep none, so their leftover entries are reclaimed here instead of waiting out
# the age rule.
LIVE_CCACHE_KEYS: 'ccache-Arch ccache-Debian13 ccache-Fedora45debug ccache-Fedora45release ccache-Windows10'
LIVE_CCACHE_KEYS: 'ccache-Arch ccache-Debian13 ccache-Fedora45debug ccache-Fedora45release ccache-Windows10 ccache-macOS14 ccache-macOS15 ccache-macOS27'
run: |
gh api --paginate "repos/$GITHUB_REPOSITORY/actions/caches?per_page=100" \
--jq '.actions_caches[] | [.id, .ref, .key, .last_accessed_at] | @tsv' >"$RUNNER_TEMP/caches.tsv"