Skip to content

Add an experiment to disable pip build isolation #37331

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
shunping merged 4 commits into from
Feb 2, 2026
+53 −0
Merged

Add an experiment to disable pip build isolation #37331

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
7b9931e
Disable build isolation for workflow tarball by default. Add an env v…
shunping Jan 16, 2026
42768bd
Apply build isolation setting to all pip install commands. Use experi…
shunping Jan 21, 2026
a6ece8d
Only search experiment in experiment options.
shunping Jan 27, 2026
7de2527
Merge branch 'master' into workflow-build-isolation
shunping Jan 28, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
39 changes: 39 additions & 0 deletions sdks/python/container/boot.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@ import (
"os/signal"
"path/filepath"
"regexp"
"slices"
"strings"
"sync"
"syscall"
Expand Down Expand Up @@ -116,6 +117,37 @@ func main() {
}
}

// The json string of pipeline options is in the following format.
// We only focus on experiments here.
//
// {
// "display_data": [
// {...},
// ],
// "options": {
// ...
// "experiments": [
// ...
// ],
// }
// }
type PipelineOptionsData struct {
Options OptionsData `json:"options"`
}

type OptionsData struct {
Experiments []string `json:"experiments"`
}

func getExperiments(options string) []string {
var opts PipelineOptionsData
err := json.Unmarshal([]byte(options), &opts)
if err != nil {
return nil
}
return opts.Options.Experiments
}

func launchSDKProcess() error {
ctx := grpcx.WriteWorkerID(context.Background(), *id)

Expand Down Expand Up @@ -155,6 +187,13 @@ func launchSDKProcess() error {
logger.Fatalf(ctx, "Failed to convert pipeline options: %v", err)
}

experiments := getExperiments(options)
Copy link
Contributor

@tvalentyn tvalentyn Feb 2, 2026 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

do you think there might be a better home for the getExperiments helper than boot.go?

Copy link
Contributor

@tvalentyn tvalentyn Feb 2, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

maybe in github.com/apache/beam/sdks/v2/go/pkg/beam/util/optionsx ?

Overtime people might add other helpers like this

beam/sdks/python/apache_beam/options/pipeline_options.py

Line 1573 in 953f70c

def lookup_experiment(self, key, default=None):
, and use it in other boot.go files.

Copy link
Collaborator Author

@shunping shunping Feb 2, 2026 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sure. Let's do that in a separate PR. We will do another round of refactoring on other boot.go in there.

Copy link
Collaborator Author

@shunping shunping Feb 2, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

On a second thought, I think we may put that into https://github.com/apache/beam/blob/master/sdks/go/container/tools/pipeline_options.go.

The parsing of pipeline options in json format seems to be only used in boot.go for different SDKs.

pipNoBuildIsolation = false
if slices.Contains(experiments, "pip_no_build_isolation") {
pipNoBuildIsolation = true
logger.Printf(ctx, "Disabled build isolation when installing packages with pip")
}

// (2) Retrieve and install the staged packages.
//
// No log.Fatalf() from here on, otherwise deferred cleanups will not be called!
Expand Down
14 changes: 14 additions & 0 deletions sdks/python/container/piputil.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,6 +32,11 @@ import (
"github.com/apache/beam/sdks/v2/go/pkg/beam/util/execx"
)

var (
// Whether to append "--no-build-isolation" flag to pip install command
pipNoBuildIsolation bool
)

const pipLogFlushInterval time.Duration = 15 * time.Second
const unrecoverableURL string = "https://beam.apache.org/documentation/sdks/python-unrecoverable-errors/index.html#pip-dependency-resolution-failures"

Expand Down Expand Up @@ -112,6 +117,9 @@ func pipInstallPackage(ctx context.Context, logger *tools.Logger, files []string
// installed if necessary. This achieves our goal outlined above.
args := []string{"-m", "pip", "install", "--no-cache-dir", "--disable-pip-version-check", "--upgrade", "--force-reinstall", "--no-deps",
filepath.Join(dir, packageSpec)}
if pipNoBuildIsolation {
args = append(args, "--no-build-isolation")
}
err := execx.ExecuteEnvWithIO(nil, os.Stdin, bufLogger, bufLogger, pythonVersion, args...)
if err != nil {
bufLogger.FlushAtError(ctx)
Expand All @@ -120,6 +128,9 @@ func pipInstallPackage(ctx context.Context, logger *tools.Logger, files []string
bufLogger.FlushAtDebug(ctx)
}
args = []string{"-m", "pip", "install", "--no-cache-dir", "--disable-pip-version-check", filepath.Join(dir, packageSpec)}
if pipNoBuildIsolation {
args = append(args, "--no-build-isolation")
}
err = execx.ExecuteEnvWithIO(nil, os.Stdin, bufLogger, bufLogger, pythonVersion, args...)
if err != nil {
bufLogger.FlushAtError(ctx)
Expand All @@ -131,6 +142,9 @@ func pipInstallPackage(ctx context.Context, logger *tools.Logger, files []string

// Case when we do not perform a forced reinstall.
args := []string{"-m", "pip", "install", "--no-cache-dir", "--disable-pip-version-check", filepath.Join(dir, packageSpec)}
if pipNoBuildIsolation {
args = append(args, "--no-build-isolation")
}
err := execx.ExecuteEnvWithIO(nil, os.Stdin, bufLogger, bufLogger, pythonVersion, args...)
if err != nil {
bufLogger.FlushAtError(ctx)
Expand Down
Loading