Skip to content

OCPBUGS-77924: cleanup orphaned boot entries in IPC #6054

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
danmanor wants to merge 1 commit into
base: main
Choose a base branch
from
+160 −9
Open

OCPBUGS-77924: cleanup orphaned boot entries in IPC #6054

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
62 changes: 58 additions & 4 deletions controllers/ipc_idle_handlers.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -151,10 +151,18 @@ func (h *IPCIdleStageHandler) cleanup(logger logr.Logger) error {
return fmt.Errorf("failed to remount sysroot: %w", err)
}

if err := h.ChrootOps.RemountBoot(); err != nil {
return fmt.Errorf("failed to remount boot: %w", err)
}

if err := h.cleanuoUnbootedStateroots(logger); err != nil {
return fmt.Errorf("failed to clean up unbooted stateroots: %w", err)
}

if err := removeOrphanedBootEntries(logger, h.ChrootOps, h.RPMOstreeClient); err != nil {
return err
}

if err := cleanupLCAWorkspace(h.ChrootOps); err != nil {
return fmt.Errorf("failed to cleanup workspace: %w", err)
}
Expand Down Expand Up @@ -199,10 +207,6 @@ func (h *IPCIdleStageHandler) cleanuoUnbootedStateroots(logger logr.Logger) erro
}
logger.Info("Stateroots to remove", "stateroots", staterootsToRemove)

if err := h.ChrootOps.RemountBoot(); err != nil {
return fmt.Errorf("failed to remount boot: %w", err)
}

if err := removeBootDirsByStaterootPrefixes(logger, h.ChrootOps, staterootsToRemove); err != nil {
return err
}
Expand Down Expand Up @@ -250,6 +254,56 @@ func removeBootDirsByStaterootPrefixes(
return nil
}

// removeOrphanedBootEntries removes directories under /boot/ostree that do not
// match any stateroot currently listed in rpm-ostree deployments.
func removeOrphanedBootEntries(
logger logr.Logger,
chrootOps ops.Ops,
rpmOstreeClient rpmostreeclient.IClient,
) error {
status, err := rpmOstreeClient.QueryStatus()
if err != nil {
return fmt.Errorf("failed to query rpm-ostree status: %w", err)
}

deployedStateroots := make(map[string]struct{})
for i := range status.Deployments {
deployedStateroots[status.Deployments[i].OSName] = struct{}{}
}

bootOstreePath := common.PathOutsideChroot("/boot/ostree")
entries, err := chrootOps.ReadDir(bootOstreePath)
if err != nil {
if chrootOps.IsNotExist(err) {
return nil
}
return fmt.Errorf("failed to list boot ostree directory %s: %w", bootOstreePath, err)
}

for _, e := range entries {
if !e.IsDir() {
continue
}
name := e.Name()
matched := false
for stateroot := range deployedStateroots {
if strings.HasPrefix(name, stateroot+"-") {
matched = true
break
}
}
if !matched {
dirPath := filepath.Join(bootOstreePath, name)
logger.Info("Removing orphaned boot entry with no matching deployment", "path", dirPath)
if err := chrootOps.RemoveAllFiles(dirPath); err != nil {
return fmt.Errorf("failed to remove orphaned boot directory %s: %w", dirPath, err)
}
}
}

return nil
}

func getStaterootsToRemove(rpmOstreeClient rpmostreeclient.IClient) ([]string, error) {
status, err := rpmOstreeClient.QueryStatus()
if err != nil {
Expand Down
107 changes: 102 additions & 5 deletions controllers/ipc_idle_handlers_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -399,6 +399,7 @@ func TestIPCIdleStageHandler_Handle(t *testing.T) {
CheckHealth = func(ctx context.Context, c client.Reader, l logr.Logger) error { return nil }

mockOps.EXPECT().RemountSysroot().Return(nil).Times(1)
mockOps.EXPECT().RemountBoot().Return(nil).Times(1)
mockRpm.EXPECT().QueryStatus().Return(nil, errors.New("rpm error")).Times(1)

res, err := h.Handle(ctx, ipc)
Expand Down Expand Up @@ -436,15 +437,20 @@ func TestIPCIdleStageHandler_Handle(t *testing.T) {
defer func() { CheckHealth = oldHC }()
CheckHealth = func(ctx context.Context, c client.Reader, l logr.Logger) error { return nil }

oldOsReadDir := osReadDir
defer func() { osReadDir = oldOsReadDir }()
osReadDir = func(name string) ([]os.DirEntry, error) { return []os.DirEntry{}, nil }

status := &rpmostreeclient.Status{
Deployments: []rpmostreeclient.Deployment{{OSName: "rhcos", Booted: true}},
}

mockOps.EXPECT().RemountSysroot().Return(nil).Times(1)
// getStaterootsToRemove + CleanupUnbootedStateroots
mockRpm.EXPECT().QueryStatus().Return(status, nil).Times(2)
// getStaterootsToRemove + removeOrphanedBootEntries + CleanupUnbootedStateroots
mockRpm.EXPECT().QueryStatus().Return(status, nil).Times(3)
mockOps.EXPECT().RemountBoot().Return(nil).Times(1)
mockOps.EXPECT().ReadDir(gomock.Any()).Return([]os.DirEntry{}, nil).Times(1)
// removeBootDirsByStaterootPrefixes + removeOrphanedBootEntries
mockOps.EXPECT().ReadDir(gomock.Any()).Return([]os.DirEntry{}, nil).Times(2)
mockRpm.EXPECT().RpmOstreeCleanup().Return(nil).Times(1)
mockOps.EXPECT().StatFile(gomock.Any()).Return(fakeFileInfo{}, nil).Times(1)
mockOps.EXPECT().RemoveAllFiles(gomock.Any()).Return(errors.New("rm failed")).Times(1)
Expand Down Expand Up @@ -501,14 +507,20 @@ func TestIPCIdleStageHandler_Handle(t *testing.T) {
defer func() { CheckHealth = oldHC }()
CheckHealth = func(ctx context.Context, c client.Reader, l logr.Logger) error { return nil }

oldOsReadDir := osReadDir
defer func() { osReadDir = oldOsReadDir }()
osReadDir = func(name string) ([]os.DirEntry, error) { return []os.DirEntry{}, nil }

status := &rpmostreeclient.Status{
Deployments: []rpmostreeclient.Deployment{{OSName: "rhcos", Booted: true}},
}

mockOps.EXPECT().RemountSysroot().Return(nil).Times(1)
mockRpm.EXPECT().QueryStatus().Return(status, nil).Times(2) // getStaterootsToRemove + CleanupUnbootedStateroots
// getStaterootsToRemove + removeOrphanedBootEntries + CleanupUnbootedStateroots
mockRpm.EXPECT().QueryStatus().Return(status, nil).Times(3)
mockOps.EXPECT().RemountBoot().Return(nil).Times(1)
mockOps.EXPECT().ReadDir(gomock.Any()).Return([]os.DirEntry{}, nil).Times(1)
// removeBootDirsByStaterootPrefixes + removeOrphanedBootEntries
mockOps.EXPECT().ReadDir(gomock.Any()).Return([]os.DirEntry{}, nil).Times(2)
mockRpm.EXPECT().RpmOstreeCleanup().Return(nil).Times(1)
mockOps.EXPECT().StatFile(gomock.Any()).Return(nil, errors.New("not exist")).Times(1)
mockOps.EXPECT().RemoveAllFiles(gomock.Any()).Times(0)
Expand Down Expand Up @@ -791,3 +803,88 @@ func TestRemoveBootDirsByStaterootPrefixes_NoBootDir(t *testing.T) {
err := removeBootDirsByStaterootPrefixes(logger, mockOps, []string{"rhcos"})
assert.NoError(t, err)
}

func TestRemoveOrphanedBootEntries(t *testing.T) {
gc := gomock.NewController(t)
defer gc.Finish()

mockOps := ops.NewMockOps(gc)
mockRpm := rpmostreeclient.NewMockIClient(gc)
logger := logr.Logger{}

bootOstreePath := common.PathOutsideChroot("/boot/ostree")

status := &rpmostreeclient.Status{
Deployments: []rpmostreeclient.Deployment{
{OSName: "rhcos_4.22.0_ec.5", Booted: true},
},
}
mockRpm.EXPECT().QueryStatus().Return(status, nil).Times(1)

entries := []os.DirEntry{
idleCleanupTestDirEntry{name: "rhcos-d1aaeead2ad0571c", isDir: true}, // orphaned, no matching deployment
idleCleanupTestDirEntry{name: "rhcos_192-168-128-80-d1aaeead2ad0571c", isDir: true}, // orphaned, no matching deployment
idleCleanupTestDirEntry{name: "rhcos_4.22.0_ec.5-d1aaeead2ad0571c", isDir: true}, // matches booted stateroot, keep
idleCleanupTestDirEntry{name: "somefile", isDir: false}, // not a dir, ignore
}
mockOps.EXPECT().ReadDir(bootOstreePath).Return(entries, nil).Times(1)

mockOps.EXPECT().RemoveAllFiles(filepath.Join(bootOstreePath, "rhcos-d1aaeead2ad0571c")).Return(nil).Times(1)
mockOps.EXPECT().RemoveAllFiles(filepath.Join(bootOstreePath, "rhcos_192-168-128-80-d1aaeead2ad0571c")).Return(nil).Times(1)

err := removeOrphanedBootEntries(logger, mockOps, mockRpm)
assert.NoError(t, err)
}

func TestRemoveOrphanedBootEntries_NoBootDir(t *testing.T) {
gc := gomock.NewController(t)
defer gc.Finish()

mockOps := ops.NewMockOps(gc)
mockRpm := rpmostreeclient.NewMockIClient(gc)
logger := logr.Logger{}

status := &rpmostreeclient.Status{
Deployments: []rpmostreeclient.Deployment{
{OSName: "rhcos", Booted: true},
},
}
mockRpm.EXPECT().QueryStatus().Return(status, nil).Times(1)

bootOstreePath := common.PathOutsideChroot("/boot/ostree")
mockOps.EXPECT().ReadDir(bootOstreePath).Return(nil, os.ErrNotExist).Times(1)
mockOps.EXPECT().IsNotExist(os.ErrNotExist).Return(true).Times(1)

err := removeOrphanedBootEntries(logger, mockOps, mockRpm)
assert.NoError(t, err)
}

func TestRemoveOrphanedBootEntries_MultipleDeployments(t *testing.T) {
gc := gomock.NewController(t)
defer gc.Finish()

mockOps := ops.NewMockOps(gc)
mockRpm := rpmostreeclient.NewMockIClient(gc)
logger := logr.Logger{}

bootOstreePath := common.PathOutsideChroot("/boot/ostree")

status := &rpmostreeclient.Status{
Deployments: []rpmostreeclient.Deployment{
{OSName: "rhcos_4.22.0", Booted: true},
{OSName: "rhcos_4.21.0", Booted: false},
},
}
mockRpm.EXPECT().QueryStatus().Return(status, nil).Times(1)

entries := []os.DirEntry{
idleCleanupTestDirEntry{name: "rhcos_4.22.0-aaa", isDir: true}, // matches deployed, keep
idleCleanupTestDirEntry{name: "rhcos_4.21.0-bbb", isDir: true}, // matches deployed, keep
idleCleanupTestDirEntry{name: "rhcos-old-ccc", isDir: true}, // orphaned, remove
}
mockOps.EXPECT().ReadDir(bootOstreePath).Return(entries, nil).Times(1)
mockOps.EXPECT().RemoveAllFiles(filepath.Join(bootOstreePath, "rhcos-old-ccc")).Return(nil).Times(1)

err := removeOrphanedBootEntries(logger, mockOps, mockRpm)
assert.NoError(t, err)
}