mesh-host/internal/apply/replace_image_test.go

package apply

import (
	"context"
	"errors"
	"strings"
	"testing"

	"git.novox.be/novox/mesh-sdk/go/declaration"
	"github.com/novox/mesh-host/internal/store"
)

// Issue 1742: the engine removed a running container and only then found the replacement image
// unobtainable, leaving the service down. The replacement image comes first.
func TestARunningContainerIsKeptWhenItsReplacementImageCannotBeObtained(t *testing.T) {
	d := parseTrusted(t, `{"declaration":1,"resources":[
	  {"id":"store","type":"container","name":"store","image":"`+pinned+`","env":{"PGDATA":"/data"}}
	]}`)
	var removed, created bool
	run := func(ctx context.Context, name string, args ...string) (string, error) {
		args = oldShape(name, args)
		switch args[0] {
		case "info":
			return "27.0\n", nil
		case "container":
			return "true\tsome-older-spec", nil
		case "image":
			return "", errors.New("no such image")
		case "pull":
			return "", errors.New("manifest unknown")
		case "rm":
			removed = true
			return "", nil
		case "run":
			created = true
			return "deadbeef\n", nil
		}
		return "", nil
	}
	_, _, err := Apply(context.Background(), archHost(t), d, store.State{}, store.OriginCarried, run, nil, nil)
	if removed || created {
		t.Fatalf("the running container was touched although its image is unobtainable (removed=%v created=%v)", removed, created)
	}
	if err == nil {
		t.Fatal("an unobtainable image failed nothing")
	}
	if !strings.Contains(err.Error(), pinned) {
		t.Errorf("the failure does not name the image %s: %v", pinned, err)
	}
}

func TestAReplacementImageIsPresentBeforeTheRunningContainerIsRemoved(t *testing.T) {
	d := parseTrusted(t, `{"declaration":1,"resources":[
	  {"id":"store","type":"container","name":"store","image":"`+pinned+`","env":{"PGDATA":"/data"}}
	]}`)
	want := containerSpec(d.Resources[0].(*declaration.Container), inputs{})
	var order []string
	created := false
	run := func(ctx context.Context, name string, args ...string) (string, error) {
		args = oldShape(name, args)
		switch args[0] {
		case "info":
			return "27.0\n", nil
		case "container":
			if created {
				return "true\t" + want, nil
			}
			return "true\tsome-older-spec", nil
		case "image":
			order = append(order, "inspect")
			if len(order) == 1 {
				return "", errors.New("no such image")
			}
			return "", nil
		case "pull":
			order = append(order, "pull")
		case "rm":
			order = append(order, "rm")
		case "run":
			order = append(order, "run")
			created = true
			return "deadbeef\n", nil
		}
		return "", nil
	}
	if _, _, err := Apply(context.Background(), archHost(t), d, store.State{}, store.OriginCarried, run, nil, nil); err != nil {
		t.Fatalf("apply failed: %v", err)
	}
	got := strings.Join(order, ",")
	if !strings.HasPrefix(got, "inspect,pull,inspect,rm,run") {
		t.Errorf("order of runtime calls = %s, want the image pulled before rm", got)
	}
}

// asideRun is a runtime whose container "store" runs from an older spec, recording the steps a replacement takes; run
// fails when failRun, and the new container reads back as declared otherwise.
func asideRun(t *testing.T, want string, failRun bool, order *[]string) Runner {
	created := false
	return func(ctx context.Context, name string, args ...string) (string, error) {
		switch args[0] {
		case "info":
			return "27.0\n", nil
		case "container":
			if created {
				return "true\t" + want, nil
			}
			return "true\tsome-older-spec", nil
		case "image":
			return "", nil
		case "rm", "rename", "stop", "start":
			*order = append(*order, strings.Join(args, " "))
			return "", nil
		case "run":
			*order = append(*order, "run")
			if failRun {
				return "", errors.New("port is already allocated")
			}
			created = true
			return "deadbeef\n", nil
		}
		return "", nil
	}
}

// A container whose replacement does not come up is put back: the running one is set aside (renamed and stopped), never
// removed before the new one runs, and started again under its name when the new one fails, the apply failing loudly.
// Removed first, the bus's own container left novox without a bus when its replacement was never made (issue 1945,
// live on 2026-10-25 at 11:23).
func TestAContainerWhoseReplacementFailsIsPutBack(t *testing.T) {
	d := parseTrusted(t, `{"declaration":1,"resources":[
	  {"id":"store","type":"container","name":"store","image":"`+pinned+`","env":{"PGDATA":"/data"}}
	]}`)
	want := containerSpec(d.Resources[0].(*declaration.Container), inputs{})
	var order []string
	_, _, err := Apply(context.Background(), archHost(t), d, store.State{}, store.OriginCarried, asideRun(t, want, true, &order), nil, nil)
	if err == nil || !strings.Contains(err.Error(), "started again") {
		t.Fatalf("a failed replacement: %v", err)
	}
	got := strings.Join(order, "; ")
	if !strings.Contains(got, "rename store store.replaced; stop store.replaced; run") ||
		!strings.Contains(got, "rename store.replaced store; start store") {
		t.Fatalf("the steps were %q", got)
	}
	if strings.Contains(got, "rm -f store;") && strings.Index(got, "rm -f store;") < strings.Index(got, "run") {
		t.Fatalf("the running container was removed before its replacement ran: %q", got)
	}
}

// A replacement that runs removes the container set aside, and only after it runs.
func TestAReplacementThatRunsRemovesTheContainerSetAside(t *testing.T) {
	d := parseTrusted(t, `{"declaration":1,"resources":[
	  {"id":"store","type":"container","name":"store","image":"`+pinned+`","env":{"PGDATA":"/data"}}
	]}`)
	want := containerSpec(d.Resources[0].(*declaration.Container), inputs{})
	var order []string
	if _, _, err := Apply(context.Background(), archHost(t), d, store.State{}, store.OriginCarried, asideRun(t, want, false, &order), nil, nil); err != nil {
		t.Fatal(err)
	}
	got := strings.Join(order, "; ")
	run, removed := strings.Index(got, "run"), strings.LastIndex(got, "rm -f store.replaced")
	if run < 0 || removed < run || strings.Contains(got, "start store") {
		t.Fatalf("the steps were %q: the set-aside container is removed after the replacement runs", got)
	}
}