felhom-controller/controller/internal/stacks/image_retention_r863_test.go

package stacks

import (
	"fmt"
	"os"
	"path/filepath"
	"strings"
	"sync"
	"testing"
	"time"

	"gitea.dooplex.hu/admin/felhom-controller/internal/dockerexec"
)

// R-863 (v0.294.0) — THE NIGHT'S EXACT SHAPE (2026-10-04, a fresh box): the household's first install is inside
// `compose up`; compose has pulled `mariadb@sha256:…` (stored untagged, `mariadb:<none>`) and not yet created the
// container; the controller's one-time image clean-up fires (3 minutes after its first start). v0.293.0 deleted
// the image — no container, installed app or undo named it — and `compose up` failed one second later.
// COMPANION RED-PROOF: drop the dockerexec.TryImageCleanup check in deleteUnkeptImages → the clean-up deletes
// sha256:MDB and the install fails ("the install failed").
func TestR863_OneTimeCleanupDuringFirstInstallKeepsThePulledImage(t *testing.T) {
	m := gateManager(t, "display_name: Book\ndeploy_fields:\n  - env_var: DOMAIN\n    type: domain\n  - env_var: SUBDOMAIN\n    type: subdomain\n    default: gapp\n")
	m.cfg.Paths.DataDir = filepath.Join(t.TempDir(), "data")
	cat := filepath.Join(m.cfg.Paths.DataDir, "catalog-cache", "templates", "bookstack")
	must(t, os.MkdirAll(cat, 0o755))
	must(t, os.WriteFile(filepath.Join(cat, "docker-compose.yml"), []byte("services:\n  db:\n    image: mariadb:11.4@sha256:mdb\n"), 0o644))

	f := &fakeImages{containers: map[string]string{}}
	var fmu sync.Mutex
	prev := imageDocker
	imageDocker = func(args ...string) (string, error) { fmu.Lock(); defer fmu.Unlock(); return f.run(args...) }
	t.Cleanup(func() { imageDocker = prev })

	var cleanupDone bool
	m.composeExecFn = func(_ string, _ map[string]string, args ...string) (string, error) {
		if len(args) == 0 || args[0] != "up" {
			return "", nil
		}
		// compose pulls by digest: the image exists, untagged, named by no container yet
		fmu.Lock()
		f.imgs = append(f.imgs, localImage{ID: "sha256:MDB", Repo: "mariadb", Tag: "<none>", Digest: "sha256:mdb", Size: "334MB"})
		fmu.Unlock()
		// the one-time clean-up fires NOW, from its own goroutine, as at 19:45:04
		res := make(chan bool, 1)
		go func() { _, done := m.RunImageRetentionOnce(); res <- done }()
		select {
		case cleanupDone = <-res:
		case <-time.After(10 * time.Second):
			return "", fmt.Errorf("the clean-up blocked")
		}
		// compose creates the container from the pulled image — if it is still there
		fmu.Lock()
		defer fmu.Unlock()
		for _, im := range f.imgs {
			if im.ID == "sha256:MDB" {
				f.containers["c-db"] = "sha256:MDB"
				return "", nil
			}
		}
		return "", fmt.Errorf("exit code 1\nstderr: Error response from daemon: No such image: mariadb@sha256:mdb")
	}
	done := make(chan bool, 1)
	m.SetDeployDoneHook(func(_ string, ok bool, _ string) { done <- ok })
	if _, err := m.DeployStack(DeployRequest{StackName: "gapp"}); err != nil {
		t.Fatal(err)
	}
	select {
	case ok := <-done:
		if !ok {
			t.Fatalf("the install failed: the clean-up deleted the image compose had just pulled (rmi %v)", f.rmi)
		}
	case <-time.After(20 * time.Second):
		t.Fatal("the deploy never ended")
	}
	if len(f.rmi) != 0 {
		t.Fatalf("the clean-up deleted during the install: %v", f.rmi)
	}
	if cleanupDone {
		t.Fatal("the clean-up reported done although it did not run — its marker would end it for good")
	}
	if _, err := os.Stat(m.imageRetentionMarker()); err == nil {
		t.Fatal("marker written for a pass that did not run")
	}
	// After the install the retried clean-up runs, and the app's image is kept (a container names it).
	if _, done := m.RunImageRetentionOnce(); !done {
		t.Fatal("the retried clean-up did not run once nothing was pulling")
	}
	if strings.Contains(strings.Join(f.rmi, ","), "sha256:MDB") {
		t.Fatal("the installed app's database image was deleted")
	}
}

// Image-pulling verbs hold the lock; others do not (a `ps` or `down` must never hold off a clean-up).
func TestR863_OnlyPullingVerbsHoldTheLock(t *testing.T) {
	for _, c := range []struct {
		args []string
		want bool
	}{
		{[]string{"up", "-d"}, true}, {[]string{"compose", "up", "-d", "--remove-orphans"}, true},
		{[]string{"pull"}, true}, {[]string{"-p", "x", "create"}, true}, {[]string{"run", "--rm", "x"}, true},
		{[]string{"ps"}, false}, {[]string{"down"}, false}, {[]string{"stop"}, false}, {[]string{"-p", "up", "down"}, false},
	} {
		if got := imagePullingForTest(c.args); got != c.want {
			t.Errorf("%v: pulling=%v, want %v", c.args, got, c.want)
		}
	}
}

func imagePullingForTest(args []string) bool { return dockerexec.ImagePulling(args) }