[PATCH v6 21/21] ci: parallelize testing on Windows
- From
Johannes Schindelin via GitGitGadget <gitgitgadget@gmail.com>
- Date
- Jan 29, 2019, 14:19 UTC
- Message-ID
- <1572444361982199fdab9c6f6b7e94383717b6c9.1548771561.git.gitgitgadget@gmail.com>
- In-Reply-To
- <pull.31.v6.git.gitgitgadget@gmail.com>
From: Johannes Schindelin <johannes.schindelin@gmx.de>
The fact that Git's test suite is implemented in Unix shell script that is as portable as we can muster, combined with the fact that Unix shell scripting is foreign to Windows (and therefore has to be emulated), results in pretty abysmal speed of the test suite on that platform, for pretty much no other reason than that language choice.
For comparison: while the Linux build & test is typically done within about 8 minutes, the Windows build & test typically lasts about 80 minutes in Azure Pipelines.
To help with that, let's use the Azure Pipeline feature where you can parallelize jobs, make jobs depend on each other, and pass artifacts between them.
The tests are distributed using the following heuristic: listing all test scripts ordered by size in descending order (as a cheap way to estimate the overall run time), every Nth script is run (where N is the total number of parallel jobs), starting at the index corresponding to the parallel job. This slicing is performed by a new function that is added to the `test-tool`.
To optimize the overall runtime of the entire Pipeline, we need to move the Windows jobs to the beginning (otherwise there would be a very decent chance for the Pipeline to be run only the Windows build, while all the parallel Windows test jobs wait for this single one).
We use Azure Pipelines Artifacts for both the minimal Git for Windows SDK as well as the built executables, as deduplication and caching close to the agents makes that really fast. For comparison: while downloading and unpacking the minimal Git for Windows SDK via PowerShell takes only one minute (down from anywhere between 2.5 to 7 when using a shallow clone), uploading it as Pipeline Artifact takes less than 30s and downloading and unpacking less than 20s (sometimes even as little as only twelve seconds).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> --- Makefile | 10 +++++ azure-pipelines.yml | 79 ++++++++++++++++++++++++++++++++++---- ci/make-test-artifacts.sh | 12 ++++++ ci/run-test-slice.sh | 17 ++++++++ t/helper/test-path-utils.c | 31 +++++++++++++++ 5 files changed, 141 insertions(+), 8 deletions(-) create mode 100755 ci/make-test-artifacts.sh create mode 100755 ci/run-test-slice.sh
diff --git a/Makefile b/Makefile index 044b4f77bd..daa318fe17 100644 --- a/Makefile +++ b/Makefile @@ -2927,6 +2927,16 @@ rpm:: @false .PHONY: rpm +artifacts-tar:: $(ALL_PROGRAMS) $(SCRIPT_LIB) $(BUILT_INS) $(OTHER_PROGRAMS) \ + GIT-BUILD-OPTIONS $(TEST_PROGRAMS) $(test_bindir_programs) \ + $(NO_INSTALL) $(MOFILES) + $(QUIET_SUBDIR0)templates $(QUIET_SUBDIR1) \ + SHELL_PATH='$(SHELL_PATH_SQ)' PERL_PATH='$(PERL_PATH_SQ)' + test -n "$(ARTIFACTS_DIRECTORY)" + mkdir -p "$(ARTIFACTS_DIRECTORY)" + $(TAR) czf "$(ARTIFACTS_DIRECTORY)/artifacts.tar.gz" $^ templates/blt/ +.PHONY: artifacts-tar + htmldocs = git-htmldocs-$(GIT_VERSION) manpages = git-manpages-$(GIT_VERSION) .PHONY: dist-doc distclean diff --git a/azure-pipelines.yml b/azure-pipelines.yml index 480e841a85..c329b7218b 100644 --- a/azure-pipelines.yml +++ b/azure-pipelines.yml @@ -3,8 +3,8 @@ resources: fetchDepth: 1 jobs: -- job: windows - displayName: Windows +- job: windows_build + displayName: Windows Build condition: succeeded() pool: Hosted timeoutInMinutes: 240 @@ -30,21 +30,84 @@ jobs: displayName: 'Download git-sdk-64-minimal' - powershell: | & git-sdk-64-minimal\usr\bin\bash.exe -lc @" - export DEVELOPER=1 - export NO_PERL=1 - export NO_SVN_TESTS=1 - export GIT_TEST_SKIP_REBASE_P=1 + ci/make-test-artifacts.sh artifacts + "@ + if (!$?) { exit(1) } + displayName: Build + env: + HOME: $(Build.SourcesDirectory) + MSYSTEM: MINGW64 + DEVELOPER: 1 + NO_PERL: 1 + - task: PublishPipelineArtifact@0 + displayName: 'Publish Pipeline Artifact: test artifacts' + inputs: + artifactName: 'windows-artifacts' + targetPath: '$(Build.SourcesDirectory)\artifacts' + - task: PublishPipelineArtifact@0 + displayName: 'Publish Pipeline Artifact: git-sdk-64-minimal' + inputs: + artifactName: 'git-sdk-64-minimal' + targetPath: '$(Build.SourcesDirectory)\git-sdk-64-minimal' + - powershell: | + if ("$GITFILESHAREPWD" -ne "" -and "$GITFILESHAREPWD" -ne "`$`(gitfileshare.pwd)") { + cmd /c rmdir "$(Build.SourcesDirectory)\test-cache" + } + displayName: 'Unmount test-cache' + condition: true + env: + GITFILESHAREPWD: $(gitfileshare.pwd) + +- job: windows_test + displayName: Windows Test + dependsOn: windows_build + condition: succeeded() + pool: Hosted + timeoutInMinutes: 240 + strategy: + parallel: 10 + steps: + - powershell: | + if ("$GITFILESHAREPWD" -ne "" -and "$GITFILESHAREPWD" -ne "`$`(gitfileshare.pwd)") { + net use s: \\gitfileshare.file.core.windows.net\test-cache "$GITFILESHAREPWD" /user:AZURE\gitfileshare /persistent:no + cmd /c mklink /d "$(Build.SourcesDirectory)\test-cache" S:\ + } + displayName: 'Mount test-cache' + env: + GITFILESHAREPWD: $(gitfileshare.pwd) + - task: DownloadPipelineArtifact@0 + displayName: 'Download Pipeline Artifact: test artifacts' + inputs: + artifactName: 'windows-artifacts' + targetPath: '$(Build.SourcesDirectory)' + - task: DownloadPipelineArtifact@0 + displayName: 'Download Pipeline Artifact: git-sdk-64-minimal' + inputs: + artifactName: 'git-sdk-64-minimal' + targetPath: '$(Build.SourcesDirectory)\git-sdk-64-minimal' + - powershell: | + & git-sdk-64-minimal\usr\bin\bash.exe -lc @" + test -f artifacts.tar.gz || { + echo No test artifacts found\; skipping >&2 + exit 0 + } + tar xf artifacts.tar.gz || exit 1 + + # Let Git ignore the SDK and the test-cache + printf '%s\n' /git-sdk-64-minimal/ /test-cache/ >>.git/info/exclude - ci/run-build-and-tests.sh || { + ci/run-test-slice.sh `$SYSTEM_JOBPOSITIONINPHASE `$SYSTEM_TOTALJOBSINPHASE || { ci/print-test-failures.sh exit 1 } "@ if (!$?) { exit(1) } - displayName: 'Build & Test' + displayName: 'Test (parallel)' env: HOME: $(Build.SourcesDirectory) MSYSTEM: MINGW64 + NO_SVN_TESTS: 1 + GIT_TEST_SKIP_REBASE_P: 1 - powershell: | if ("$GITFILESHAREPWD" -ne "" -and "$GITFILESHAREPWD" -ne "`$`(gitfileshare.pwd)") { cmd /c rmdir "$(Build.SourcesDirectory)\test-cache" diff --git a/ci/make-test-artifacts.sh b/ci/make-test-artifacts.sh new file mode 100755 index 0000000000..646967481f --- /dev/null +++ b/ci/make-test-artifacts.sh @@ -0,0 +1,12 @@ +#!/bin/sh +# +# Build Git and store artifacts for testing +# + +mkdir -p "$1" # in case ci/lib.sh decides to quit early + +. ${0%/*}/lib.sh + +make artifacts-tar ARTIFACTS_DIRECTORY="$1" + +check_unignored_build_artifacts diff --git a/ci/run-test-slice.sh b/ci/run-test-slice.sh new file mode 100755 index 0000000000..f8c2c3106a --- /dev/null +++ b/ci/run-test-slice.sh @@ -0,0 +1,17 @@ +#!/bin/sh +# +# Test Git in parallel +# + +. ${0%/*}/lib.sh + +case "$CI_OS_NAME" in +windows*) cmd //c mklink //j t\\.prove "$(cygpath -aw "$cache_dir/.prove")";; +*) ln -s "$cache_dir/.prove" t/.prove;; +esac + +make --quiet -C t T="$(cd t && + ./helper/test-tool path-utils slice-tests "$1" "$2" t[0-9]*.sh | + tr '\n' ' ')" + +check_unignored_build_artifacts diff --git a/t/helper/test-path-utils.c b/t/helper/test-path-utils.c index 6efde6f5ba..5d543ad21f 100644 --- a/t/helper/test-path-utils.c +++ b/t/helper/test-path-utils.c @@ -177,6 +177,14 @@ static int is_dotgitmodules(const char *path) return is_hfs_dotgitmodules(path) || is_ntfs_dotgitmodules(path); } +static int cmp_by_st_size(const void *a, const void *b) +{ + intptr_t x = (intptr_t)((struct string_list_item *)a)->util; + intptr_t y = (intptr_t)((struct string_list_item *)b)->util; + + return x > y ? -1 : (x < y ? +1 : 0); +} + int cmd__path_utils(int argc, const char **argv) { if (argc == 3 && !strcmp(argv[1], "normalize_path_copy")) { @@ -324,6 +332,29 @@ int cmd__path_utils(int argc, const char **argv) return 0; } + if (argc > 5 && !strcmp(argv[1], "slice-tests")) { + int res = 0; + long offset, stride, i; + struct string_list list = STRING_LIST_INIT_NODUP; + struct stat st; + + offset = strtol(argv[2], NULL, 10); + stride = strtol(argv[3], NULL, 10); + if (stride < 1) + stride = 1; + for (i = 4; i < argc; i++) + if (stat(argv[i], &st)) + res = error_errno("Cannot stat '%s'", argv[i]); + else + string_list_append(&list, argv[i])->util = + (void *)(intptr_t)st.st_size; + QSORT(list.items, list.nr, cmp_by_st_size); + for (i = offset; i < list.nr; i+= stride) + printf("%s\n", list.items[i].string); + + return !!res; + } + fprintf(stderr, "%s: unknown function name: %s\n", argv[0], argv[1] ? argv[1] : "(there was none)"); return 1;
-- gitgitgadget