Use only the necessary files in the cache keys. (#387)

pkg/commands/add.go

@@ -19,12 +19,13 @@ package commands
 import (
 	"path/filepath"
 
+	"github.com/moby/buildkit/frontend/dockerfile/instructions"
+
 	"github.com/GoogleContainerTools/kaniko/pkg/dockerfile"
 
 	"github.com/google/go-containerregistry/pkg/v1"
 
 	"github.com/GoogleContainerTools/kaniko/pkg/util"
-	"github.com/moby/buildkit/frontend/dockerfile/instructions"
 	"github.com/sirupsen/logrus"
 )
 
@@ -44,18 +45,13 @@ type AddCommand struct {
 // 	2. If <src> is a local tar archive:
 // 		-If <src> is a local tar archive, it is unpacked at the dest, as 'tar -x' would
 func (a *AddCommand) ExecuteCommand(config *v1.Config, buildArgs *dockerfile.BuildArgs) error {
-	// First, resolve any environment replacement
 	replacementEnvs := buildArgs.ReplacementEnvs(config.Env)
-	resolvedEnvs, err := util.ResolveEnvironmentReplacementList(a.cmd.SourcesAndDest, replacementEnvs, true)
+
-	if err != nil {
+	srcs, dest, err := resolveEnvAndWildcards(a.cmd.SourcesAndDest, a.buildcontext, replacementEnvs)
-		return err
-	}
-	dest := resolvedEnvs[len(resolvedEnvs)-1]
-	// Resolve wildcards and get a list of resolved sources
-	srcs, err := util.ResolveSources(resolvedEnvs, a.buildcontext)
 	if err != nil {
 		return err
 	}
+
 	var unresolvedSrcs []string
 	// If any of the sources are local tar archives:
 	// 	1. Unpack them to the specified destination
@@ -94,6 +90,7 @@ func (a *AddCommand) ExecuteCommand(config *v1.Config, buildArgs *dockerfile.Bui
 		},
 		buildcontext: a.buildcontext,
 	}
+
 	if err := copyCmd.ExecuteCommand(config, buildArgs); err != nil {
 		return err
 	}
@@ -111,6 +108,26 @@ func (a *AddCommand) String() string {
 	return a.cmd.String()
 }
 
-func (a *AddCommand) UsesContext() bool {
+func (a *AddCommand) FilesUsedFromContext(config *v1.Config, buildArgs *dockerfile.BuildArgs) ([]string, error) {
-	return true
+	replacementEnvs := buildArgs.ReplacementEnvs(config.Env)
+
+	srcs, _, err := resolveEnvAndWildcards(a.cmd.SourcesAndDest, a.buildcontext, replacementEnvs)
+	if err != nil {
+		return nil, err
+	}
+
+	files := []string{}
+	for _, src := range srcs {
+		if util.IsSrcRemoteFileURL(src) {
+			continue
+		}
+		if util.IsFileLocalTarArchive(src) {
+			continue
+		}
+		fullPath := filepath.Join(a.buildcontext, src)
+		files = append(files, fullPath)
+	}
+
+	logrus.Infof("Using files from context: %v", files)
+	return files, nil
 }

pkg/commands/base_command.go

@@ -16,19 +16,23 @@ limitations under the License.
 
 package commands
 
+import (
+	"github.com/GoogleContainerTools/kaniko/pkg/dockerfile"
+	"github.com/google/go-containerregistry/pkg/v1"
+)
+
 type BaseCommand struct {
-	cache       bool
+	cache bool
-	usesContext bool
 }
 
 func (b *BaseCommand) CacheCommand() bool {
 	return b.cache
 }
 
-func (b *BaseCommand) UsesContext() bool {
-	return b.usesContext
-}
-
 func (b *BaseCommand) FilesToSnapshot() []string {
 	return []string{}
 }
+
+func (b *BaseCommand) FilesUsedFromContext(_ *v1.Config, _ *dockerfile.BuildArgs) ([]string, error) {
+	return []string{}, nil
+}

pkg/commands/commands.go

@@ -39,7 +39,7 @@ type DockerCommand interface {
 	CacheCommand() bool
 
 	// Return true if this command depends on the build context.
-	UsesContext() bool
+	FilesUsedFromContext(*v1.Config, *dockerfile.BuildArgs) ([]string, error)
 }
 
 func GetCommand(cmd instructions.Command, buildcontext string) (DockerCommand, error) {

pkg/commands/copy.go

@@ -20,12 +20,14 @@ import (
 	"os"
 	"path/filepath"
 
+	"github.com/moby/buildkit/frontend/dockerfile/instructions"
+	"github.com/sirupsen/logrus"
+
 	"github.com/GoogleContainerTools/kaniko/pkg/constants"
 
 	"github.com/GoogleContainerTools/kaniko/pkg/dockerfile"
 	"github.com/GoogleContainerTools/kaniko/pkg/util"
 	"github.com/google/go-containerregistry/pkg/v1"
-	"github.com/moby/buildkit/frontend/dockerfile/instructions"
 )
 
 type CopyCommand struct {
@@ -40,18 +42,14 @@ func (c *CopyCommand) ExecuteCommand(config *v1.Config, buildArgs *dockerfile.Bu
 	if c.cmd.From != "" {
 		c.buildcontext = filepath.Join(constants.KanikoDir, c.cmd.From)
 	}
+
 	replacementEnvs := buildArgs.ReplacementEnvs(config.Env)
-	// First, resolve any environment replacement
+
-	resolvedEnvs, err := util.ResolveEnvironmentReplacementList(c.cmd.SourcesAndDest, replacementEnvs, true)
+	srcs, dest, err := resolveEnvAndWildcards(c.cmd.SourcesAndDest, c.buildcontext, replacementEnvs)
-	if err != nil {
-		return err
-	}
-	dest := resolvedEnvs[len(resolvedEnvs)-1]
-	// Resolve wildcards and get a list of resolved sources
-	srcs, err := util.ResolveSources(resolvedEnvs, c.buildcontext)
 	if err != nil {
 		return err
 	}
+
 	// For each source, iterate through and copy it over
 	for _, src := range srcs {
 		fullPath := filepath.Join(c.buildcontext, src)
@@ -94,6 +92,18 @@ func (c *CopyCommand) ExecuteCommand(config *v1.Config, buildArgs *dockerfile.Bu
 	return nil
 }
 
+func resolveEnvAndWildcards(sd instructions.SourcesAndDest, buildcontext string, envs []string) ([]string, string, error) {
+	// First, resolve any environment replacement
+	resolvedEnvs, err := util.ResolveEnvironmentReplacementList(sd, envs, true)
+	if err != nil {
+		return nil, "", err
+	}
+	dest := resolvedEnvs[len(resolvedEnvs)-1]
+	// Resolve wildcards and get a list of resolved sources
+	srcs, err := util.ResolveSources(resolvedEnvs, buildcontext)
+	return srcs, dest, err
+}
+
 // FilesToSnapshot should return an empty array if still nil; no files were changed
 func (c *CopyCommand) FilesToSnapshot() []string {
 	return c.snapshotFiles
@@ -104,6 +114,23 @@ func (c *CopyCommand) String() string {
 	return c.cmd.String()
 }
 
-func (c *CopyCommand) UsesContext() bool {
+func (c *CopyCommand) FilesUsedFromContext(config *v1.Config, buildArgs *dockerfile.BuildArgs) ([]string, error) {
-	return true
+	// We don't use the context if we're performing a copy --from.
+	if c.cmd.From != "" {
+		return nil, nil
+	}
+
+	replacementEnvs := buildArgs.ReplacementEnvs(config.Env)
+	srcs, _, err := resolveEnvAndWildcards(c.cmd.SourcesAndDest, c.buildcontext, replacementEnvs)
+	if err != nil {
+		return nil, err
+	}
+
+	files := []string{}
+	for _, src := range srcs {
+		fullPath := filepath.Join(c.buildcontext, src)
+		files = append(files, fullPath)
+	}
+	logrus.Infof("Using files from context: %v", files)
+	return files, nil
 }

pkg/executor/build.go

@@ -125,10 +125,6 @@ func (s *stageBuilder) build() error {
 
 	// Set the initial cache key to be the base image digest, the build args and the SrcContext.
 	compositeKey := NewCompositeCache(s.baseImageDigest)
-	contextHash, err := HashDir(s.opts.SrcContext)
-	if err != nil {
-		return err
-	}
 	compositeKey.AddKey(s.opts.BuildArgs...)
 
 	cmds := []commands.DockerCommand{}
@@ -148,8 +144,16 @@ func (s *stageBuilder) build() error {
 
 		// Add the next command to the cache key.
 		compositeKey.AddKey(command.String())
-		if command.UsesContext() {
+
-			compositeKey.AddKey(contextHash)
+		// If the command uses files from the context, add them.
+		files, err := command.FilesUsedFromContext(&s.cf.Config, args)
+		if err != nil {
+			return err
+		}
+		for _, f := range files {
+			if err := compositeKey.AddPath(f); err != nil {
+				return err
+			}
 		}
 		logrus.Info(command.String())
 
@@ -172,7 +176,7 @@ func (s *stageBuilder) build() error {
 		if err := command.ExecuteCommand(&s.cf.Config, args); err != nil {
 			return err
 		}
-		files := command.FilesToSnapshot()
+		files = command.FilesToSnapshot()
 		var contents []byte
 
 		if !s.shouldTakeSnapshot(index, files) {

pkg/executor/composite_cache.go

@@ -53,6 +53,32 @@ func (s *CompositeCache) Hash() (string, error) {
 	return util.SHA256(strings.NewReader(s.Key()))
 }
 
+func (s *CompositeCache) AddPath(p string) error {
+	sha := sha256.New()
+	fi, err := os.Lstat(p)
+	if err != nil {
+		return err
+	}
+	if fi.Mode().IsDir() {
+		k, err := HashDir(p)
+		if err != nil {
+			return err
+		}
+		s.keys = append(s.keys, k)
+		return nil
+	}
+	fh, err := util.CacheHasher()(p)
+	if err != nil {
+		return err
+	}
+	if _, err := sha.Write([]byte(fh)); err != nil {
+		return err
+	}
+
+	s.keys = append(s.keys, string(sha.Sum(nil)))
+	return nil
+}
+
 // HashDir returns a hash of the directory.
 func HashDir(p string) (string, error) {
 	sha := sha256.New()