Files
kaniko/pkg/executor/build.go
T
Christie Wilson 607af5f7a6 Always snapshot files in COPY and RUN commands
Kaniko uses mtime (as well as file contents and other attributes) to
determine if files have changed. COPY and ADD commands should _always_
update the mtime, because they actually overwrite the files. However it
turns out that the mtime can lag, so kaniko would sometimes add a new
layer when using COPY or ADD on a file, and sometimes would not. This
leads to a non-deterministic number of layers.

To fix this, we have updated the kaniko commands to be more
authoritative in declaring when they have changed a file (e.g. WORKDIR
will now only create the directory when it doesn't exist) and we will
trust those files and _always_ add them, instead of only adding them if
they haven't changed.

It is possible for RUN commands to also change the filesystem, in which
case kaniko has no choice but to look at the filesystem to determine
what has changed. For this case we have added a call to `sync` however
we still cannot guarantee that sometimes the mtime will not lag, causing the
number of layers to be non-deterministic. However when I tried to cause
this behaviour with the RUN command, I couldn't.

This changes the snapshotting logic a bit; before this change, the last
command of the last stage in a Dockerfile would always scan the whole
file system and ignore the files returned by the kaniko command. Instead
we will now trust those files and assume that the snapshotting
performed by previous commands will be adequate.

Docker itself seems to rely on the storage driver to determine when
files have changed and so doesn't have to deal with these problems
directly.

An alternative implementation would use `inotify` to track which files
have changed. However that would mean watching every file in the
filesystem, and adding new watches as files are added. Not only is there
a limit on the number of files that can be watched, but according to the
man pages a) this can take a significant amount of time b) there is
complication around when events arrive (e.g. by the time they arrive,
the files may have changed) and lastly c) events can be lost, which
would mean we'd run into this non-deterministic behaviour again anyway.

Fixes #251
2018-08-23 18:23:39 -07:00

237 lines
7.0 KiB
Go

/*
Copyright 2018 Google LLC
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/
package executor
import (
"bytes"
"fmt"
"io"
"io/ioutil"
"os"
"path/filepath"
"strconv"
"github.com/google/go-containerregistry/pkg/name"
"github.com/google/go-containerregistry/pkg/v1"
"github.com/google/go-containerregistry/pkg/v1/mutate"
"github.com/google/go-containerregistry/pkg/v1/tarball"
"github.com/moby/buildkit/frontend/dockerfile/instructions"
"github.com/sirupsen/logrus"
"github.com/GoogleContainerTools/kaniko/pkg/commands"
"github.com/GoogleContainerTools/kaniko/pkg/constants"
"github.com/GoogleContainerTools/kaniko/pkg/dockerfile"
"github.com/GoogleContainerTools/kaniko/pkg/options"
"github.com/GoogleContainerTools/kaniko/pkg/snapshot"
"github.com/GoogleContainerTools/kaniko/pkg/util"
)
func DoBuild(opts *options.KanikoOptions) (v1.Image, error) {
// Parse dockerfile and unpack base image to root
stages, err := dockerfile.Stages(opts.DockerfilePath, opts.Target)
if err != nil {
return nil, err
}
hasher, err := getHasher(opts.SnapshotMode)
if err != nil {
return nil, err
}
for index, stage := range stages {
finalStage := finalStage(index, opts.Target, stages)
// Unpack file system to root
sourceImage, err := util.RetrieveSourceImage(index, opts.BuildArgs, stages)
if err != nil {
return nil, err
}
if err := util.GetFSFromImage(constants.RootDir, sourceImage); err != nil {
return nil, err
}
l := snapshot.NewLayeredMap(hasher)
snapshotter := snapshot.NewSnapshotter(l, constants.RootDir)
// Take initial snapshot
if err := snapshotter.Init(); err != nil {
return nil, err
}
imageConfig, err := util.RetrieveConfigFile(sourceImage)
if err != nil {
return nil, err
}
if err := resolveOnBuild(&stage, &imageConfig.Config); err != nil {
return nil, err
}
buildArgs := dockerfile.NewBuildArgs(opts.BuildArgs)
for index, cmd := range stage.Commands {
finalCmd := index == len(stage.Commands)-1
dockerCommand, err := commands.GetCommand(cmd, opts.SrcContext)
if err != nil {
return nil, err
}
if dockerCommand == nil {
continue
}
if err := dockerCommand.ExecuteCommand(&imageConfig.Config, buildArgs); err != nil {
return nil, err
}
snapshotFiles := dockerCommand.FilesToSnapshot()
var contents []byte
// If this is an intermediate stage, we only snapshot for the last command and we
// want to snapshot the entire filesystem since we aren't tracking what was changed
// by previous commands.
if !finalStage {
if finalCmd {
contents, err = snapshotter.TakeSnapshotFS()
}
} else {
// If we are in single snapshot mode, we only take a snapshot once, after all
// commands have completed.
if opts.SingleSnapshot {
if finalCmd {
contents, err = snapshotter.TakeSnapshotFS()
}
} else {
// Otherwise, in the final stage we take a snapshot at each command. If we know
// the files that were changed, we'll snapshot those explicitly, otherwise we'll
// check if anything in the filesystem changed.
if snapshotFiles != nil {
contents, err = snapshotter.TakeSnapshot(snapshotFiles)
} else {
contents, err = snapshotter.TakeSnapshotFS()
}
}
}
if err != nil {
return nil, fmt.Errorf("Error taking snapshot of files for command %s: %s", dockerCommand, err)
}
util.MoveVolumeWhitelistToWhitelist()
if contents == nil {
logrus.Info("No files were changed, appending empty layer to config. No layer added to image.")
continue
}
// Append the layer to the image
opener := func() (io.ReadCloser, error) {
return ioutil.NopCloser(bytes.NewReader(contents)), nil
}
layer, err := tarball.LayerFromOpener(opener)
if err != nil {
return nil, err
}
sourceImage, err = mutate.Append(sourceImage,
mutate.Addendum{
Layer: layer,
History: v1.History{
Author: constants.Author,
CreatedBy: dockerCommand.CreatedBy(),
},
},
)
if err != nil {
return nil, err
}
}
sourceImage, err = mutate.Config(sourceImage, imageConfig.Config)
if err != nil {
return nil, err
}
if finalStage {
if opts.Reproducible {
sourceImage, err = mutate.Canonical(sourceImage)
if err != nil {
return nil, err
}
}
return sourceImage, nil
}
if dockerfile.SaveStage(index, stages) {
if err := saveStageAsTarball(index, sourceImage); err != nil {
return nil, err
}
if err := extractImageToDependecyDir(index, sourceImage); err != nil {
return nil, err
}
}
// Delete the filesystem
if err := util.DeleteFilesystem(); err != nil {
return nil, err
}
}
return nil, err
}
func finalStage(index int, target string, stages []instructions.Stage) bool {
if index == len(stages)-1 {
return true
}
if target == "" {
return false
}
return target == stages[index].Name
}
func extractImageToDependecyDir(index int, image v1.Image) error {
dependencyDir := filepath.Join(constants.KanikoDir, strconv.Itoa(index))
if err := os.MkdirAll(dependencyDir, 0755); err != nil {
return err
}
logrus.Infof("trying to extract to %s", dependencyDir)
return util.GetFSFromImage(dependencyDir, image)
}
func saveStageAsTarball(stageIndex int, image v1.Image) error {
destRef, err := name.NewTag("temp/tag", name.WeakValidation)
if err != nil {
return err
}
if err := os.MkdirAll(constants.KanikoIntermediateStagesDir, 0750); err != nil {
return err
}
tarPath := filepath.Join(constants.KanikoIntermediateStagesDir, strconv.Itoa(stageIndex))
logrus.Infof("Storing source image from stage %d at path %s", stageIndex, tarPath)
return tarball.WriteToFile(tarPath, destRef, image, nil)
}
func getHasher(snapshotMode string) (func(string) (string, error), error) {
if snapshotMode == constants.SnapshotModeTime {
logrus.Info("Only file modification time will be considered when snapshotting")
return util.MtimeHasher(), nil
}
if snapshotMode == constants.SnapshotModeFull {
return util.Hasher(), nil
}
return nil, fmt.Errorf("%s is not a valid snapshot mode", snapshotMode)
}
func resolveOnBuild(stage *instructions.Stage, config *v1.Config) error {
if config.OnBuild == nil {
return nil
}
// Otherwise, parse into commands
cmds, err := dockerfile.ParseCommands(config.OnBuild)
if err != nil {
return err
}
// Append to the beginning of the commands in the stage
stage.Commands = append(cmds, stage.Commands...)
logrus.Infof("Executing %v build triggers", len(cmds))
// Blank out the Onbuild command list for this image
config.OnBuild = nil
return nil
}