c8d: implement missing image delete logic

Ports over all the previous image delete logic, such as:
- Introduce `prune` and `force` flags
- Introduce the concept of hard and soft image delete conflics, which represent:
  - image referenced in multiple tags (soft conflict)
  - image being used by a stopped container (soft conflict)
  - image being used by a running container (hard conflict)
- Implement delete logic such as:
  - if deleting by reference, and there are other references to the same image, just
    delete the passed reference
  - if deleting by reference, and there is only 1 reference and the image is being used
    by a running container, throw an error if !force, or delete the reference and create
    a dangling reference otherwise
  - if deleting by imageID, and force is true, remove all tags (otherwise soft conflict)
  - if imageID, check if stopped container is using the image (soft conflict), and
    delete anyway if force
  - if imageID was passed in, check if running container is using the image (hard conflict)
  - if `prune` is true, and the image being deleted has dangling parents, remove them

This commit also implements logic to get image parents in c8d by comparing shared layers.

Signed-off-by: Laura Brehm <laurabrehm@hey.com>
This commit is contained in:
Laura Brehm 2023-04-25 22:43:51 +01:00
parent f117aef2ea
commit cad97135b3
No known key found for this signature in database
GPG key ID: 526E3FC49260D47A
2 changed files with 340 additions and 20 deletions

View file

@ -127,3 +127,71 @@ func isRootfsChildOf(child ocispec.RootFS, parent ocispec.RootFS) bool {
return true
}
// parents returns a slice of image IDs whose entire rootfs contents match,
// in order, the childs first layers, excluding images with the exact same
// rootfs.
//
// Called from image_delete.go to prune dangling parents.
func (i *ImageService) parents(ctx context.Context, id image.ID) ([]imageWithRootfs, error) {
target, err := i.resolveDescriptor(ctx, id.String())
if err != nil {
return nil, errors.Wrap(err, "failed to get child image")
}
cs := i.client.ContentStore()
allPlatforms, err := containerdimages.Platforms(ctx, cs, target)
if err != nil {
return nil, errdefs.System(errors.Wrap(err, "failed to list platforms supported by image"))
}
var childRootFS []ocispec.RootFS
for _, platform := range allPlatforms {
rootfs, err := platformRootfs(ctx, cs, target, platform)
if err != nil {
if cerrdefs.IsNotFound(err) {
continue
}
return nil, errdefs.System(errors.Wrap(err, "failed to get platform-specific rootfs"))
}
childRootFS = append(childRootFS, rootfs)
}
imgs, err := i.client.ImageService().List(ctx)
if err != nil {
return nil, errdefs.System(errors.Wrap(err, "failed to list all images"))
}
var parents []imageWithRootfs
for _, img := range imgs {
nextImage:
for _, platform := range allPlatforms {
rootfs, err := platformRootfs(ctx, cs, img.Target, platform)
if err != nil {
if cerrdefs.IsNotFound(err) {
continue
}
return nil, errdefs.System(errors.Wrap(err, "failed to get platform-specific rootfs"))
}
for _, childRoot := range childRootFS {
if isRootfsChildOf(childRoot, rootfs) {
parents = append(parents, imageWithRootfs{
img: img,
rootfs: rootfs,
})
break nextImage
}
}
}
}
return parents, nil
}
type imageWithRootfs struct {
img containerdimages.Image
rootfs ocispec.RootFS
}

View file

@ -2,10 +2,16 @@ package containerd
import (
"context"
"fmt"
"sort"
"strings"
"github.com/containerd/containerd/images"
"github.com/docker/distribution/reference"
"github.com/docker/docker/api/types"
"github.com/docker/docker/container"
"github.com/docker/docker/image"
"github.com/docker/docker/pkg/stringid"
"github.com/opencontainers/go-digest"
ocispec "github.com/opencontainers/image-spec/specs-go/v1"
"github.com/sirupsen/logrus"
@ -30,8 +36,6 @@ import (
// are divided into two categories grouped by their severity:
//
// Hard Conflict:
// - a pull or build using the image.
// - any descendant image.
// - any running container using the image.
//
// Soft Conflict:
@ -45,8 +49,6 @@ import (
// meaning any delete conflicts will cause the image to not be deleted and the
// conflict will not be reported.
//
// TODO(thaJeztah): implement ImageDelete "force" options; see https://github.com/moby/moby/issues/43850
// TODO(thaJeztah): implement ImageDelete "prune" options; see https://github.com/moby/moby/issues/43849
// TODO(thaJeztah): image delete should send prometheus counters; see https://github.com/moby/moby/issues/45268
func (i *ImageService) ImageDelete(ctx context.Context, imageRef string, force, prune bool) ([]types.ImageDeleteResponseItem, error) {
parsedRef, err := reference.ParseNormalizedNamed(imageRef)
@ -59,28 +61,278 @@ func (i *ImageService) ImageDelete(ctx context.Context, imageRef string, force,
return nil, err
}
possiblyDeletedConfigs := map[digest.Digest]struct{}{}
if err := i.walkPresentChildren(ctx, img.Target, func(_ context.Context, d ocispec.Descriptor) {
if images.IsConfigType(d.MediaType) {
possiblyDeletedConfigs[d.Digest] = struct{}{}
}
}); err != nil {
return nil, err
imgID := image.ID(img.Target.Digest)
if isImageIDPrefix(imgID.String(), imageRef) {
return i.deleteAll(ctx, img, force, prune)
}
err = i.client.ImageService().Delete(ctx, img.Name, images.SynchronousDelete())
singleRef, err := i.isSingleReference(ctx, img)
if err != nil {
return nil, err
}
// Workaround for: https://github.com/moby/buildkit/issues/3797
if err := i.unleaseSnapshotsFromDeletedConfigs(context.Background(), possiblyDeletedConfigs); err != nil {
logrus.WithError(err).Warn("failed to unlease snapshots")
if !singleRef {
err := i.client.ImageService().Delete(ctx, img.Name)
if err != nil {
return nil, err
}
i.LogImageEvent(imgID.String(), imgID.String(), "untag")
records := []types.ImageDeleteResponseItem{{Untagged: reference.FamiliarString(reference.TagNameOnly(parsedRef))}}
return records, nil
}
imgID := string(img.Target.Digest)
i.LogImageEvent(imgID, imgID, "untag")
i.LogImageEvent(imgID, imgID, "delete")
using := func(c *container.Container) bool {
return c.ImageID == imgID
}
ctr := i.containers.First(using)
if ctr != nil {
if !force {
// If we removed the repository reference then
// this image would remain "dangling" and since
// we really want to avoid that the client must
// explicitly force its removal.
refString := reference.FamiliarString(reference.TagNameOnly(parsedRef))
err := &imageDeleteConflict{
reference: refString,
used: true,
message: fmt.Sprintf("container %s is using its referenced image %s",
stringid.TruncateID(ctr.ID),
stringid.TruncateID(imgID.String())),
}
return nil, err
}
return []types.ImageDeleteResponseItem{{Untagged: reference.FamiliarString(parsedRef)}}, nil
err := i.softImageDelete(ctx, img)
if err != nil {
return nil, err
}
i.LogImageEvent(imgID.String(), imgID.String(), "untag")
records := []types.ImageDeleteResponseItem{{Untagged: reference.FamiliarString(reference.TagNameOnly(parsedRef))}}
return records, nil
}
return i.deleteAll(ctx, img, force, prune)
}
// deleteAll deletes the image from the daemon, and if prune is true,
// also deletes dangling parents if there is no conflict in doing so.
// Parent images are removed quietly, and if there is any issue/conflict
// it is logged but does not halt execution/an error is not returned.
func (i *ImageService) deleteAll(ctx context.Context, img images.Image, force, prune bool) ([]types.ImageDeleteResponseItem, error) {
var records []types.ImageDeleteResponseItem
// Workaround for: https://github.com/moby/buildkit/issues/3797
possiblyDeletedConfigs := map[digest.Digest]struct{}{}
err := i.walkPresentChildren(ctx, img.Target, func(_ context.Context, d ocispec.Descriptor) {
if images.IsConfigType(d.MediaType) {
possiblyDeletedConfigs[d.Digest] = struct{}{}
}
})
if err != nil {
return nil, err
}
defer func() {
if err := i.unleaseSnapshotsFromDeletedConfigs(context.Background(), possiblyDeletedConfigs); err != nil {
logrus.WithError(err).Warn("failed to unlease snapshots")
}
}()
imgID := img.Target.Digest.String()
var parents []imageWithRootfs
if prune {
parents, err = i.parents(ctx, image.ID(imgID))
if err != nil {
logrus.WithError(err).Warn("failed to get image parents")
}
sortParentsByAffinity(parents)
}
imageRefs, err := i.client.ImageService().List(ctx, "target.digest=="+imgID)
if err != nil {
return nil, err
}
for _, imageRef := range imageRefs {
if err := i.imageDeleteHelper(ctx, imageRef, &records, force); err != nil {
return records, err
}
}
i.LogImageEvent(imgID, imgID, "delete")
records = append(records, types.ImageDeleteResponseItem{Deleted: imgID})
for _, parent := range parents {
if !isDanglingImage(parent.img) {
break
}
err = i.imageDeleteHelper(ctx, parent.img, &records, false)
if err != nil {
logrus.WithError(err).Warn("failed to remove image parent")
break
}
parentID := parent.img.Target.Digest.String()
i.LogImageEvent(parentID, parentID, "delete")
records = append(records, types.ImageDeleteResponseItem{Deleted: parentID})
}
return records, nil
}
// isImageIDPrefix returns whether the given
// possiblePrefix is a prefix of the given imageID.
func isImageIDPrefix(imageID, possiblePrefix string) bool {
if strings.HasPrefix(imageID, possiblePrefix) {
return true
}
if i := strings.IndexRune(imageID, ':'); i >= 0 {
return strings.HasPrefix(imageID[i+1:], possiblePrefix)
}
return false
}
func sortParentsByAffinity(parents []imageWithRootfs) {
sort.Slice(parents, func(i, j int) bool {
lenRootfsI := len(parents[i].rootfs.DiffIDs)
lenRootfsJ := len(parents[j].rootfs.DiffIDs)
if lenRootfsI == lenRootfsJ {
return isDanglingImage(parents[i].img)
}
return lenRootfsI > lenRootfsJ
})
}
// isSingleReference returns true if there are no other images in the
// daemon targeting the same content as `img` that are not dangling.
func (i *ImageService) isSingleReference(ctx context.Context, img images.Image) (bool, error) {
refs, err := i.client.ImageService().List(ctx, "target.digest=="+img.Target.Digest.String())
if err != nil {
return false, err
}
for _, ref := range refs {
if !isDanglingImage(ref) && ref.Name != img.Name {
return false, nil
}
}
return true, nil
}
type conflictType int
const (
conflictRunningContainer conflictType = 1 << iota
conflictActiveReference
conflictStoppedContainer
conflictHard = conflictRunningContainer
conflictSoft = conflictActiveReference | conflictStoppedContainer
)
// imageDeleteHelper attempts to delete the given image from this daemon.
// If the image has any hard delete conflicts (running containers using
// the image) then it cannot be deleted. If the image has any soft delete
// conflicts (any tags/digests referencing the image or any stopped container
// using the image) then it can only be deleted if force is true. Any deleted
// images and untagged references are appended to the given records. If any
// error or conflict is encountered, it will be returned immediately without
// deleting the image.
func (i *ImageService) imageDeleteHelper(ctx context.Context, img images.Image, records *[]types.ImageDeleteResponseItem, force bool) error {
// First, determine if this image has any conflicts. Ignore soft conflicts
// if force is true.
c := conflictHard
if !force {
c |= conflictSoft
}
imgID := image.ID(img.Target.Digest)
err := i.checkImageDeleteConflict(ctx, imgID, c)
if err != nil {
return err
}
untaggedRef, err := reference.ParseAnyReference(img.Name)
if err != nil {
return err
}
err = i.client.ImageService().Delete(ctx, img.Name, images.SynchronousDelete())
if err != nil {
return err
}
i.LogImageEvent(imgID.String(), imgID.String(), "untag")
*records = append(*records, types.ImageDeleteResponseItem{Untagged: reference.FamiliarString(untaggedRef)})
return nil
}
// ImageDeleteConflict holds a soft or hard conflict and associated
// error. A hard conflict represents a running container using the
// image, while a soft conflict is any tags/digests referencing the
// given image or any stopped container using the image.
// Implements the error interface.
type imageDeleteConflict struct {
hard bool
used bool
reference string
message string
}
func (idc *imageDeleteConflict) Error() string {
var forceMsg string
if idc.hard {
forceMsg = "cannot be forced"
} else {
forceMsg = "must be forced"
}
return fmt.Sprintf("conflict: unable to delete %s (%s) - %s", idc.reference, forceMsg, idc.message)
}
func (imageDeleteConflict) Conflict() {}
// checkImageDeleteConflict returns a conflict representing
// any issue preventing deletion of the given image ID, and
// nil if there are none. It takes a bitmask representing a
// filter for which conflict types the caller cares about,
// and will only check for these conflict types.
func (i *ImageService) checkImageDeleteConflict(ctx context.Context, imgID image.ID, mask conflictType) error {
if mask&conflictRunningContainer != 0 {
running := func(c *container.Container) bool {
return c.ImageID == imgID && c.IsRunning()
}
if ctr := i.containers.First(running); ctr != nil {
return &imageDeleteConflict{
reference: stringid.TruncateID(imgID.String()),
hard: true,
used: true,
message: fmt.Sprintf("image is being used by running container %s", stringid.TruncateID(ctr.ID)),
}
}
}
if mask&conflictStoppedContainer != 0 {
stopped := func(c *container.Container) bool {
return !c.IsRunning() && c.ImageID == imgID
}
if ctr := i.containers.First(stopped); ctr != nil {
return &imageDeleteConflict{
reference: stringid.TruncateID(imgID.String()),
used: true,
message: fmt.Sprintf("image is being used by stopped container %s", stringid.TruncateID(ctr.ID)),
}
}
}
if mask&conflictActiveReference != 0 {
refs, err := i.client.ImageService().List(ctx, "target.digest=="+imgID.String())
if err != nil {
return err
}
if len(refs) > 1 {
return &imageDeleteConflict{
reference: stringid.TruncateID(imgID.String()),
message: "image is referenced in multiple repositories",
}
}
}
return nil
}