Merge 73ca2e7426 into 236ba5690e

This attempts to mitigate some of the sources for issues described in
https://github.com/GoogleContainerTools/kaniko/issues/1697

The potential problems that this might address is when something is
writing to the filesystem during it's deletion but is otherwise safe
to be removed anyway

MAINTAINERS

@@ -1,2 +0,0 @@
-Jerome Ju <jeromeju@google.com>
-Quan Zhang <zhangquan@google.com>

README.md

@@ -1,3 +1,11 @@
+# 🧊 This project is archived and no longer developed or maintained. 🧊
+
+The code remains available for historic purposes.
+
+The README as of the archival date remains unchanged below for historic purposes.
+
+-----
+
 # kaniko - Build Images In Kubernetes
 
 ## 🚨NOTE: kaniko is not an officially supported Google product🚨
@@ -844,6 +852,11 @@ Cache timeout in hours. Defaults to two weeks.
 
 Set this flag to clean the filesystem at the end of the build.
 
+#### --retry-cleanup
+
+Set this flag to retry cleanup when it fails
+The failure can happen when a volume is being mounted into an image and receives writes during the cleanup
+
 #### Flag `--compressed-caching`
 
 Set this to false in order to prevent tar compression for cached layers. This

cmd/executor/cmd/root.go

@@ -259,6 +259,7 @@ func addKanikoOptionsFlags() {
 	RootCmd.PersistentFlags().BoolVarP(&opts.Cache, "cache", "", false, "Use cache when building image")
 	RootCmd.PersistentFlags().BoolVarP(&opts.CompressedCaching, "compressed-caching", "", true, "Compress the cached layers. Decreases build time, but increases memory usage.")
 	RootCmd.PersistentFlags().BoolVarP(&opts.Cleanup, "cleanup", "", false, "Clean the filesystem at the end")
+	RootCmd.PersistentFlags().BoolVarP(&opts.RetryCleanup, "retry-cleanup", "", false, "Add retries to filesystem cleanup between stages and after the build is done")
 	RootCmd.PersistentFlags().DurationVarP(&opts.CacheTTL, "cache-ttl", "", time.Hour*336, "Cache timeout, requires value and unit of duration -> ex: 6h. Defaults to two weeks.")
 	RootCmd.PersistentFlags().VarP(&opts.InsecureRegistries, "insecure-registry", "", "Insecure registry using plain HTTP to push and pull. Set it repeatedly for multiple registries.")
 	RootCmd.PersistentFlags().VarP(&opts.SkipTLSVerifyRegistries, "skip-tls-verify-registry", "", "Insecure registry ignoring TLS verify to push and pull. Set it repeatedly for multiple registries.")

pkg/config/options.go

@@ -82,6 +82,7 @@ type KanikoOptions struct {
 	NoPushCache              bool
 	Cache                    bool
 	Cleanup                  bool
+	RetryCleanup             bool
 	CompressedCaching        bool
 	IgnoreVarRun             bool
 	SkipUnusedStages         bool

pkg/executor/build.go

@@ -787,7 +787,7 @@ func DoBuild(opts *config.KanikoOptions) (v1.Image, error) {
 				}
 			}
 			if opts.Cleanup {
-				if err = util.DeleteFilesystem(); err != nil {
+				if err = util.DeleteFilesystem(opts.RetryCleanup); err != nil {
 					return nil, err
 				}
 			}
@@ -819,7 +819,7 @@ func DoBuild(opts *config.KanikoOptions) (v1.Image, error) {
 		}
 
 		// Delete the filesystem
-		if err := util.DeleteFilesystem(); err != nil {
+		if err := util.DeleteFilesystem(opts.RetryCleanup); err != nil {
 			return nil, errors.Wrap(err, fmt.Sprintf("deleting file system after stage %d", index))
 		}
 	}

pkg/util/fs_util.go

@@ -230,7 +230,7 @@ func GetFSFromLayers(root string, layers []v1.Layer, opts ...FSOpt) ([]string, e
 }
 
 // DeleteFilesystem deletes the extracted image file system
-func DeleteFilesystem() error {
+func DeleteFilesystem(retry bool) error {
 	logrus.Info("Deleting filesystem...")
 	return filepath.Walk(config.RootDir, func(path string, info os.FileInfo, err error) error {
 		if err != nil {
@@ -256,7 +256,19 @@ func DeleteFilesystem() error {
 		if path == config.RootDir {
 			return nil
 		}
-		return os.RemoveAll(path)
+		if retry {
+			for i := 0; i < 3; i++ { // Retry 3 times
+				err := os.RemoveAll(path)
+				if err == nil {
+					return nil
+				}
+				logrus.Warnf("Error deleting %s, retrying (%d/3): %v", path, i+1, err)
+				time.Sleep(1 * time.Second) // Wait before retrying
+			}
+			return fmt.Errorf("failed to delete %s after 3 retries", path)
+		} else {
+			return os.RemoveAll(path)
+		}
 	})
 }