Merge pull request 'cron task to cleanup dangling container images with version sha256:*' (#4698) from earl-warren/forgejo:wip-container-cleanup into forgejo

Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/4698 Reviewed-by: Gusted <gusted@noreply.codeberg.org>
2024-11-08 18:04:14 +01:00 · 2024-08-22 08:09:19 +00:00 · 2024-08-22 08:09:19 +00:00 · b670f111ff
commit b670f111ff
parent 801ae21200 f19f31ac73
3 changed files with 383 additions and 0 deletions
--- a/services/packages/container/cleanup.go
+++ b/services/packages/container/cleanup.go
@ -21,6 +21,9 @@ func Cleanup(ctx context.Context, olderThan time.Duration) error {
 	if err := cleanupExpiredBlobUploads(ctx, olderThan); err != nil {
 		return err
 	}
+	if err := CleanupSHA256(ctx, olderThan); err != nil {
+		return err
+	}
 	return cleanupExpiredUploadedBlobs(ctx, olderThan)
 }

--- a/services/packages/container/cleanup_sha256.go
+++ b/services/packages/container/cleanup_sha256.go
@ -0,0 +1,142 @@
+// Copyright 2024 The Forgejo Authors. All rights reserved.
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+package container
+
+import (
+	"context"
+	"strings"
+	"time"
+
+	"code.gitea.io/gitea/models/db"
+	"code.gitea.io/gitea/models/packages"
+	"code.gitea.io/gitea/modules/json"
+	"code.gitea.io/gitea/modules/log"
+	container_module "code.gitea.io/gitea/modules/packages/container"
+)
+
+var (
+	SHA256BatchSize = 500
+	SHA256Log       = "cleanup dangling images with a sha256:* version"
+	SHA256LogStart  = "Start to " + SHA256Log
+	SHA256LogFinish = "Finished to " + SHA256Log
+)
+
+func CleanupSHA256(ctx context.Context, olderThan time.Duration) error {
+	log.Info(SHA256LogStart)
+	err := cleanupSHA256(ctx, olderThan)
+	log.Info(SHA256LogFinish)
+	return err
+}
+
+func cleanupSHA256(outerCtx context.Context, olderThan time.Duration) error {
+	ctx, committer, err := db.TxContext(outerCtx)
+	if err != nil {
+		return err
+	}
+	defer committer.Close()
+
+	foundAtLeastOneSHA256 := false
+	shaToVersionID := make(map[string]int64, 100)
+	knownSHA := make(map[string]any, 100)
+
+	log.Debug("Look for all package_version.version that start with sha256:")
+
+	old := time.Now().Add(-olderThan).Unix()
+
+	// Iterate over all container versions in ascending order and store
+	// in shaToVersionID all versions with a sha256: prefix. If an index
+	// manifest is found, the sha256: digest it references are removed
+	// from shaToVersionID. If the sha256: digest found in an index
+	// manifest is not already in shaToVersionID, it is stored in
+	// knownSHA to be dealt with later.
+	//
+	// Although it is theoretically possible that a sha256: is uploaded
+	// after the index manifest that references it, this is not the
+	// normal order of operations. First the sha256: version is uploaded
+	// and then the index manifest. When the iteration completes,
+	// knownSHA will therefore be empty most of the time and
+	// shaToVersionID will only contain unreferenced sha256: versions.
+	if err := db.GetEngine(ctx).
+		Select("`package_version`.`id`, `package_version`.`lower_version`, `package_version`.`metadata_json`").
+		Join("INNER", "`package`", "`package`.`id` = `package_version`.`package_id`").
+		Where("`package`.`type` = ? AND `package_version`.`created_unix` < ?", packages.TypeContainer, old).
+		OrderBy("`package_version`.`id` ASC").
+		Iterate(new(packages.PackageVersion), func(_ int, bean any) error {
+			v := bean.(*packages.PackageVersion)
+			if strings.HasPrefix(v.LowerVersion, "sha256:") {
+				shaToVersionID[v.LowerVersion] = v.ID
+				foundAtLeastOneSHA256 = true
+			} else if strings.Contains(v.MetadataJSON, `"manifests":[{`) {
+				var metadata container_module.Metadata
+				if err := json.Unmarshal([]byte(v.MetadataJSON), &metadata); err != nil {
+					log.Error("package_version.id = %d package_version.metadata_json %s is not a JSON string containing valid metadata. It was ignored but it is an inconsistency in the database that should be looked at. %v", v.ID, v.MetadataJSON, err)
+					return nil
+				}
+				for _, manifest := range metadata.Manifests {
+					if _, ok := shaToVersionID[manifest.Digest]; ok {
+						delete(shaToVersionID, manifest.Digest)
+					} else {
+						knownSHA[manifest.Digest] = true
+					}
+				}
+			}
+			return nil
+		}); err != nil {
+		return err
+	}
+
+	for sha := range knownSHA {
+		delete(shaToVersionID, sha)
+	}
+
+	if len(shaToVersionID) == 0 {
+		if foundAtLeastOneSHA256 {
+			log.Debug("All container images with a version matching sha256:* are referenced by an index manifest")
+		} else {
+			log.Debug("There are no container images with a version matching sha256:*")
+		}
+		log.Info("Nothing to cleanup")
+		return nil
+	}
+
+	found := len(shaToVersionID)
+
+	log.Warn("%d container image(s) with a version matching sha256:* are not referenced by an index manifest", found)
+
+	log.Debug("Deleting unreferenced image versions from `package_version`, `package_file` and `package_property` (%d at a time)", SHA256BatchSize)
+
+	packageVersionIDs := make([]int64, 0, SHA256BatchSize)
+	for _, id := range shaToVersionID {
+		packageVersionIDs = append(packageVersionIDs, id)
+	}
+
+	for len(packageVersionIDs) > 0 {
+		upper := min(len(packageVersionIDs), SHA256BatchSize)
+		versionIDs := packageVersionIDs[0:upper]
+
+		var packageFileIDs []int64
+		if err := db.GetEngine(ctx).Select("id").Table("package_file").In("version_id", versionIDs).Find(&packageFileIDs); err != nil {
+			return err
+		}
+		log.Info("Removing %d entries from `package_file` and `package_property`", len(packageFileIDs))
+		if _, err := db.GetEngine(ctx).In("id", packageFileIDs).Delete(&packages.PackageFile{}); err != nil {
+			return err
+		}
+		if _, err := db.GetEngine(ctx).In("ref_id", packageFileIDs).And("ref_type = ?", packages.PropertyTypeFile).Delete(&packages.PackageProperty{}); err != nil {
+			return err
+		}
+
+		log.Info("Removing %d entries from `package_version` and `package_property`", upper)
+		if _, err := db.GetEngine(ctx).In("id", versionIDs).Delete(&packages.PackageVersion{}); err != nil {
+			return err
+		}
+		if _, err := db.GetEngine(ctx).In("ref_id", versionIDs).And("ref_type = ?", packages.PropertyTypeVersion).Delete(&packages.PackageProperty{}); err != nil {
+			return err
+		}
+
+		packageVersionIDs = packageVersionIDs[upper:]
+	}
+
+	return committer.Commit()
+}
--- a/tests/integration/api_packages_container_cleanup_sha256_test.go
+++ b/tests/integration/api_packages_container_cleanup_sha256_test.go
@ -0,0 +1,238 @@
+// Copyright 2024 The Forgejo Authors. All rights reserved.
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+package integration
+
+import (
+	"bytes"
+	"encoding/base64"
+	"fmt"
+	"net/http"
+	"strings"
+	"testing"
+	"time"
+
+	"code.gitea.io/gitea/models/db"
+	packages_model "code.gitea.io/gitea/models/packages"
+	"code.gitea.io/gitea/models/unittest"
+	user_model "code.gitea.io/gitea/models/user"
+	"code.gitea.io/gitea/modules/log"
+	packages_module "code.gitea.io/gitea/modules/packages"
+	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/test"
+	packages_cleanup "code.gitea.io/gitea/services/packages/cleanup"
+	packages_container "code.gitea.io/gitea/services/packages/container"
+	"code.gitea.io/gitea/tests"
+
+	oci "github.com/opencontainers/image-spec/specs-go/v1"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestPackagesContainerCleanupSHA256(t *testing.T) {
+	defer tests.PrepareTestEnv(t, 1)()
+	defer test.MockVariableValue(&setting.Packages.Storage.Type, setting.LocalStorageType)()
+	defer test.MockVariableValue(&packages_container.SHA256BatchSize, 1)()
+
+	ctx := db.DefaultContext
+
+	user := unittest.AssertExistsAndLoadBean(t, &user_model.User{ID: 2})
+
+	cleanupAndCheckLogs := func(t *testing.T, expected ...string) {
+		t.Helper()
+		logChecker, cleanup := test.NewLogChecker(log.DEFAULT, log.TRACE)
+		logChecker.Filter(expected...)
+		logChecker.StopMark(packages_container.SHA256LogFinish)
+		defer cleanup()
+
+		require.NoError(t, packages_cleanup.CleanupExpiredData(ctx, -1*time.Hour))
+
+		logFiltered, logStopped := logChecker.Check(5 * time.Second)
+		assert.True(t, logStopped)
+		filtered := make([]bool, 0, len(expected))
+		for range expected {
+			filtered = append(filtered, true)
+		}
+		assert.EqualValues(t, filtered, logFiltered, expected)
+	}
+
+	userToken := ""
+
+	t.Run("Authenticate", func(t *testing.T) {
+		type TokenResponse struct {
+			Token string `json:"token"`
+		}
+
+		authenticate := []string{`Bearer realm="` + setting.AppURL + `v2/token",service="container_registry",scope="*"`}
+
+		t.Run("User", func(t *testing.T) {
+			req := NewRequest(t, "GET", fmt.Sprintf("%sv2", setting.AppURL))
+			resp := MakeRequest(t, req, http.StatusUnauthorized)
+
+			assert.ElementsMatch(t, authenticate, resp.Header().Values("WWW-Authenticate"))
+
+			req = NewRequest(t, "GET", fmt.Sprintf("%sv2/token", setting.AppURL)).
+				AddBasicAuth(user.Name)
+			resp = MakeRequest(t, req, http.StatusOK)
+
+			tokenResponse := &TokenResponse{}
+			DecodeJSON(t, resp, &tokenResponse)
+
+			assert.NotEmpty(t, tokenResponse.Token)
+
+			userToken = fmt.Sprintf("Bearer %s", tokenResponse.Token)
+
+			req = NewRequest(t, "GET", fmt.Sprintf("%sv2", setting.AppURL)).
+				AddTokenAuth(userToken)
+			MakeRequest(t, req, http.StatusOK)
+		})
+	})
+
+	image := "test"
+	multiTag := "multi"
+
+	url := fmt.Sprintf("%sv2/%s/%s", setting.AppURL, user.Name, image)
+
+	blobDigest := "sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4"
+	sha256ManifestDigest := "sha256:4305f5f5572b9a426b88909b036e52ee3cf3d7b9c1b01fac840e90747f56623d"
+	indexManifestDigest := "sha256:b992f98104ab25f60d78368a674ce6f6a49741f4e32729e8496067ed06174e9b"
+
+	uploadSHA256Version := func(t *testing.T) {
+		t.Helper()
+
+		blobContent, _ := base64.StdEncoding.DecodeString(`H4sIAAAJbogA/2IYBaNgFIxYAAgAAP//Lq+17wAEAAA=`)
+
+		req := NewRequestWithBody(t, "POST", fmt.Sprintf("%s/blobs/uploads?digest=%s", url, blobDigest), bytes.NewReader(blobContent)).
+			AddTokenAuth(userToken)
+		resp := MakeRequest(t, req, http.StatusCreated)
+
+		assert.Equal(t, fmt.Sprintf("/v2/%s/%s/blobs/%s", user.Name, image, blobDigest), resp.Header().Get("Location"))
+		assert.Equal(t, blobDigest, resp.Header().Get("Docker-Content-Digest"))
+
+		configDigest := "sha256:4607e093bec406eaadb6f3a340f63400c9d3a7038680744c406903766b938f0d"
+		configContent := `{"architecture":"amd64","config":{"Env":["PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin"],"Cmd":["/true"],"ArgsEscaped":true,"Image":"sha256:9bd8b88dc68b80cffe126cc820e4b52c6e558eb3b37680bfee8e5f3ed7b8c257"},"container":"b89fe92a887d55c0961f02bdfbfd8ac3ddf66167db374770d2d9e9fab3311510","container_config":{"Hostname":"b89fe92a887d","Env":["PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin"],"Cmd":["/bin/sh","-c","#(nop) ","CMD [\"/true\"]"],"ArgsEscaped":true,"Image":"sha256:9bd8b88dc68b80cffe126cc820e4b52c6e558eb3b37680bfee8e5f3ed7b8c257"},"created":"2022-01-01T00:00:00.000000000Z","docker_version":"20.10.12","history":[{"created":"2022-01-01T00:00:00.000000000Z","created_by":"/bin/sh -c #(nop) COPY file:0e7589b0c800daaf6fa460d2677101e4676dd9491980210cb345480e513f3602 in /true "},{"created":"2022-01-01T00:00:00.000000001Z","created_by":"/bin/sh -c #(nop)  CMD [\"/true\"]","empty_layer":true}],"os":"linux","rootfs":{"type":"layers","diff_ids":["sha256:0ff3b91bdf21ecdf2f2f3d4372c2098a14dbe06cd678e8f0a85fd4902d00e2e2"]}}`
+
+		req = NewRequestWithBody(t, "POST", fmt.Sprintf("%s/blobs/uploads?digest=%s", url, configDigest), strings.NewReader(configContent)).
+			AddTokenAuth(userToken)
+		MakeRequest(t, req, http.StatusCreated)
+
+		sha256ManifestContent := `{"schemaVersion":2,"mediaType":"` + oci.MediaTypeImageManifest + `","config":{"mediaType":"application/vnd.docker.container.image.v1+json","digest":"sha256:4607e093bec406eaadb6f3a340f63400c9d3a7038680744c406903766b938f0d","size":1069},"layers":[{"mediaType":"application/vnd.docker.image.rootfs.diff.tar.gzip","digest":"sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4","size":32}]}`
+		req = NewRequestWithBody(t, "PUT", fmt.Sprintf("%s/manifests/%s", url, sha256ManifestDigest), strings.NewReader(sha256ManifestContent)).
+			AddTokenAuth(userToken).
+			SetHeader("Content-Type", oci.MediaTypeImageManifest)
+		resp = MakeRequest(t, req, http.StatusCreated)
+
+		assert.Equal(t, sha256ManifestDigest, resp.Header().Get("Docker-Content-Digest"))
+
+		req = NewRequest(t, "HEAD", fmt.Sprintf("%s/manifests/%s", url, sha256ManifestDigest)).
+			AddTokenAuth(userToken)
+		resp = MakeRequest(t, req, http.StatusOK)
+
+		assert.Equal(t, fmt.Sprintf("%d", len(sha256ManifestContent)), resp.Header().Get("Content-Length"))
+		assert.Equal(t, sha256ManifestDigest, resp.Header().Get("Docker-Content-Digest"))
+	}
+
+	uploadIndexManifest := func(t *testing.T) {
+		indexManifestContent := `{"schemaVersion":2,"mediaType":"` + oci.MediaTypeImageIndex + `","manifests":[{"mediaType":"application/vnd.docker.distribution.manifest.v2+json","digest":"` + sha256ManifestDigest + `","platform":{"os":"linux","architecture":"arm","variant":"v7"}}]}`
+		req := NewRequestWithBody(t, "PUT", fmt.Sprintf("%s/manifests/%s", url, multiTag), strings.NewReader(indexManifestContent)).
+			AddTokenAuth(userToken).
+			SetHeader("Content-Type", oci.MediaTypeImageIndex)
+		resp := MakeRequest(t, req, http.StatusCreated)
+
+		assert.Equal(t, indexManifestDigest, resp.Header().Get("Docker-Content-Digest"))
+	}
+
+	assertImageExists := func(t *testing.T, manifestDigest, blobDigest string) {
+		req := NewRequest(t, "HEAD", fmt.Sprintf("%s/manifests/%s", url, manifestDigest)).
+			AddTokenAuth(userToken)
+		MakeRequest(t, req, http.StatusOK)
+
+		req = NewRequest(t, "HEAD", fmt.Sprintf("%s/blobs/%s", url, blobDigest)).
+			AddTokenAuth(userToken)
+		MakeRequest(t, req, http.StatusOK)
+	}
+
+	assertImageNotExists := func(t *testing.T, manifestDigest, blobDigest string) {
+		req := NewRequest(t, "HEAD", fmt.Sprintf("%s/manifests/%s", url, manifestDigest)).
+			AddTokenAuth(userToken)
+		MakeRequest(t, req, http.StatusNotFound)
+
+		req = NewRequest(t, "HEAD", fmt.Sprintf("%s/blobs/%s", url, blobDigest)).
+			AddTokenAuth(userToken)
+		MakeRequest(t, req, http.StatusNotFound)
+	}
+
+	assertImageDeleted := func(t *testing.T, image, manifestDigest, blobDigest string, cleanup func()) {
+		t.Helper()
+		packageVersion := unittest.AssertExistsAndLoadBean(t, &packages_model.PackageVersion{Version: manifestDigest})
+		packageFile := unittest.AssertExistsAndLoadBean(t, &packages_model.PackageFile{VersionID: packageVersion.ID})
+		unittest.AssertExistsAndLoadBean(t, &packages_model.PackageProperty{RefID: packageFile.ID, RefType: packages_model.PropertyTypeVersion})
+		packageBlob := unittest.AssertExistsAndLoadBean(t, &packages_model.PackageBlob{ID: packageFile.BlobID})
+		contentStore := packages_module.NewContentStore()
+		require.NoError(t, contentStore.Has(packages_module.BlobHash256Key(packageBlob.HashSHA256)))
+
+		assertImageExists(t, manifestDigest, blobDigest)
+
+		cleanup()
+
+		assertImageNotExists(t, manifestDigest, blobDigest)
+
+		unittest.AssertNotExistsBean(t, &packages_model.PackageVersion{Version: manifestDigest})
+		unittest.AssertNotExistsBean(t, &packages_model.PackageFile{VersionID: packageVersion.ID})
+		unittest.AssertNotExistsBean(t, &packages_model.PackageProperty{RefID: packageFile.ID, RefType: packages_model.PropertyTypeVersion})
+		unittest.AssertNotExistsBean(t, &packages_model.PackageBlob{ID: packageFile.BlobID})
+		assert.Error(t, contentStore.Has(packages_module.BlobHash256Key(packageBlob.HashSHA256)))
+	}
+
+	assertImageAndPackageDeleted := func(t *testing.T, image, manifestDigest, blobDigest string, cleanup func()) {
+		t.Helper()
+		unittest.AssertExistsAndLoadBean(t, &packages_model.Package{Name: image})
+		assertImageDeleted(t, image, manifestDigest, blobDigest, cleanup)
+		unittest.AssertNotExistsBean(t, &packages_model.Package{Name: image})
+	}
+
+	t.Run("Nothing to look at", func(t *testing.T) {
+		cleanupAndCheckLogs(t, "There are no container images with a version matching sha256:*")
+	})
+
+	uploadSHA256Version(t)
+
+	t.Run("Dangling image found", func(t *testing.T) {
+		assertImageAndPackageDeleted(t, image, sha256ManifestDigest, blobDigest, func() {
+			cleanupAndCheckLogs(t,
+				"Removing 3 entries from `package_file` and `package_property`",
+				"Removing 1 entries from `package_version` and `package_property`",
+			)
+		})
+	})
+
+	uploadSHA256Version(t)
+	uploadIndexManifest(t)
+
+	t.Run("Corrupted index manifest metadata is ignored", func(t *testing.T) {
+		assertImageExists(t, sha256ManifestDigest, blobDigest)
+		_, err := db.GetEngine(ctx).Table("package_version").Where("version = ?", multiTag).Update(&packages_model.PackageVersion{MetadataJSON: `corrupted "manifests":[{ bad`})
+		require.NoError(t, err)
+
+		// do not expect the package to be deleted because it contains
+		// corrupted metadata that prevents that from happening
+		assertImageDeleted(t, image, sha256ManifestDigest, blobDigest, func() {
+			cleanupAndCheckLogs(t,
+				"Removing 3 entries from `package_file` and `package_property`",
+				"Removing 1 entries from `package_version` and `package_property`",
+				"is not a JSON string containing valid metadata",
+			)
+		})
+	})
+
+	uploadSHA256Version(t)
+	uploadIndexManifest(t)
+
+	t.Run("Image found but referenced", func(t *testing.T) {
+		assertImageExists(t, sha256ManifestDigest, blobDigest)
+		cleanupAndCheckLogs(t,
+			"All container images with a version matching sha256:* are referenced by an index manifest",
+		)
+		assertImageExists(t, sha256ManifestDigest, blobDigest)
+	})
+}