From e26eae499c7b21df27c170a3f174fbda5bd2c6b5 Mon Sep 17 00:00:00 2001
From: Matthias Bertschy <matthias.bertschy@gmail.com>
Date: Mon, 26 Aug 2024 18:30:15 +0200
Subject: [PATCH] warns if cleanup finds files bigger than 30MB

Signed-off-by: Matthias Bertschy <matthias.bertschy@gmail.com>
---
 main.go                     |  2 +-
 pkg/cleanup/cleanup.go      | 24 +++++++++++++++---------
 pkg/cleanup/cleanup_test.go |  3 ++-
 3 files changed, 18 insertions(+), 11 deletions(-)

diff --git a/main.go b/main.go
index c9d75cca5..b8f6fffeb 100644
--- a/main.go
+++ b/main.go
@@ -81,7 +81,7 @@ func main() {
 		file.DefaultStorageRoot,
 		intervalDuration,
 		kubernetesAPI)
-	go cleanupHandler.StartCleanupTask()
+	go cleanupHandler.StartCleanupTask(ctx)
 
 	logger.L().Info("APIServer started")
 	code := cli.Run(cmd)
diff --git a/pkg/cleanup/cleanup.go b/pkg/cleanup/cleanup.go
index 00a102129..bd255baeb 100644
--- a/pkg/cleanup/cleanup.go
+++ b/pkg/cleanup/cleanup.go
@@ -2,6 +2,7 @@ package cleanup
 
 import (
 	"bytes"
+	"context"
 	"fmt"
 	"os"
 	"path/filepath"
@@ -19,6 +20,10 @@ import (
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 )
 
+const (
+	MinSizeToReport = 30 * 1024 * 1024 // 30MB
+)
+
 type TypeCleanupHandlerFunc func(kind, path string, metadata *metav1.ObjectMeta, resourceMaps ResourceMaps) bool
 
 var resourceKindToHandler = map[string]TypeCleanupHandlerFunc{
@@ -65,7 +70,7 @@ func NewResourcesCleanupHandler(appFs afero.Fs, root string, interval time.Durat
 	}
 }
 
-func (h *ResourcesCleanupHandler) StartCleanupTask() {
+func (h *ResourcesCleanupHandler) StartCleanupTask(ctx context.Context) {
 	for {
 		logger.L().Info("started cleanup task", helpers.String("interval", h.interval.String()))
 		var err error
@@ -76,7 +81,6 @@ func (h *ResourcesCleanupHandler) StartCleanupTask() {
 			continue
 		}
 
-		var size int64
 		for resourceKind, handler := range resourceKindToHandler {
 			v1beta1ApiVersionPath := filepath.Join(h.root, softwarecomposition.GroupName, resourceKind)
 			exists, _ := afero.DirExists(h.appFs, v1beta1ApiVersionPath)
@@ -88,9 +92,13 @@ func (h *ResourcesCleanupHandler) StartCleanupTask() {
 					return err
 				}
 
-				// sum all files in path
-				if !info.IsDir() {
-					size += info.Size()
+				// skip directories
+				if info.IsDir() {
+					return nil
+				}
+
+				if size := info.Size(); size > MinSizeToReport {
+					logger.L().Ctx(ctx).Warning("large file detected, you may want to truncate it", helpers.String("path", path), helpers.String("size", fmt.Sprintf("%d bytes", size)))
 				}
 
 				// FIXME: migrate to gob files - to remove after some time
@@ -127,8 +135,8 @@ func (h *ResourcesCleanupHandler) StartCleanupTask() {
 					}
 				}
 
-				// skip directories and files that are not metadata files
-				if info.IsDir() || !file.IsMetadataFile(path) {
+				// skip files that are not metadata files
+				if !file.IsMetadataFile(path) {
 					return nil
 				}
 
@@ -157,8 +165,6 @@ func (h *ResourcesCleanupHandler) StartCleanupTask() {
 			}
 		}
 
-		logger.L().Info("storage size before cleanup", helpers.String("path", h.root), helpers.String("size", fmt.Sprintf("%d bytes", size)))
-
 		if h.interval == 0 {
 			break
 		}
diff --git a/pkg/cleanup/cleanup_test.go b/pkg/cleanup/cleanup_test.go
index 481ae30e0..36d0bcc5b 100644
--- a/pkg/cleanup/cleanup_test.go
+++ b/pkg/cleanup/cleanup_test.go
@@ -2,6 +2,7 @@ package cleanup
 
 import (
 	"archive/zip"
+	"context"
 	"encoding/json"
 	"fmt"
 	"io"
@@ -50,7 +51,7 @@ func TestCleanupTask(t *testing.T) {
 		fetcher:    &ResourcesFetchMock{},
 		deleteFunc: deleteFunc,
 	}
-	handler.StartCleanupTask()
+	handler.StartCleanupTask(context.TODO())
 
 	expectedFilesToDelete := []string{
 		"/data/spdx.softwarecomposition.kubescape.io/applicationactivities/gadget/gadget-daemonset-gadget-0d7c-fd3c.g",