Scrub runsc error messages

Removed "error" and "failed to" prefix that don't add value
from messages. Adjusted a few other messages.  In particular,
when the container fail to start, the message returned is easier
for humans to read:

$ docker run --rm --runtime=runsc alpine foobar
docker: Error response from daemon: OCI runtime start failed: <path> did not terminate sucessfully: starting container: starting root container [foobar]: starting sandbox: searching for executable "foobar", cwd: "/", $PATH="/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin": no such file or directory

Closes #77

PiperOrigin-RevId: 230022798
Change-Id: I83339017c70dae09e4f9f8e0ea2e554c4d5d5cd1

29 files changed, 211 insertions, 201 deletions

diff --git a/runsc/boot/controller.go b/runsc/boot/controller.go
index 36e9d2c6b..989f49388 100644
--- a/runsc/boot/controller.go
+++ b/runsc/boot/controller.go
@@ -163,7 +163,7 @@ func (cm *containerManager) StartRoot(cid *string, _ *struct{}) error {
 	// Tell the root container to start and wait for the result.
 	cm.startChan <- struct{}{}
 	if err := <-cm.startResultChan; err != nil {
-		return fmt.Errorf("failed to start sandbox: %v", err)
+		return fmt.Errorf("starting sandbox: %v", err)
 	}
 	return nil
 }
@@ -319,7 +319,7 @@ func (cm *containerManager) Restore(o *RestoreOpts, _ *struct{}) error {
 
 	p, err := createPlatform(cm.l.conf, int(deviceFile.Fd()))
 	if err != nil {
-		return fmt.Errorf("error creating platform: %v", err)
+		return fmt.Errorf("creating platform: %v", err)
 	}
 	k := &kernel.Kernel{
 		Platform: p,
@@ -330,14 +330,14 @@ func (cm *containerManager) Restore(o *RestoreOpts, _ *struct{}) error {
 	fds := &fdDispenser{fds: cm.l.goferFDs}
 	renv, err := createRestoreEnvironment(cm.l.spec, cm.l.conf, fds)
 	if err != nil {
-		return fmt.Errorf("error creating RestoreEnvironment: %v", err)
+		return fmt.Errorf("creating RestoreEnvironment: %v", err)
 	}
 	fs.SetRestoreEnvironment(*renv)
 
 	// Prepare to load from the state file.
 	networkStack, err := newEmptyNetworkStack(cm.l.conf, k)
 	if err != nil {
-		return fmt.Errorf("failed to create network: %v", err)
+		return fmt.Errorf("creating network: %v", err)
 	}
 	if eps, ok := networkStack.(*epsocket.Stack); ok {
 		stack.StackFromEnv = eps.Stack // FIXME
@@ -347,7 +347,7 @@ func (cm *containerManager) Restore(o *RestoreOpts, _ *struct{}) error {
 		return err
 	}
 	if info.Size() == 0 {
-		return fmt.Errorf("error file was empty")
+		return fmt.Errorf("file cannot be empty")
 	}
 
 	// Load the state.
@@ -385,7 +385,7 @@ func (cm *containerManager) Restore(o *RestoreOpts, _ *struct{}) error {
 	// Tell the root container to start and wait for the result.
 	cm.startChan <- struct{}{}
 	if err := <-cm.startResultChan; err != nil {
-		return fmt.Errorf("failed to start sandbox: %v", err)
+		return fmt.Errorf("starting sandbox: %v", err)
 	}
 
 	return nil
diff --git a/runsc/boot/fs.go b/runsc/boot/fs.go
index e0c8291ac..5c5e650ca 100644
--- a/runsc/boot/fs.go
+++ b/runsc/boot/fs.go
@@ -98,11 +98,11 @@ func setupRootContainerFS(userCtx context.Context, rootCtx context.Context, spec
 	fds := &fdDispenser{fds: goferFDs}
 	rootInode, err := createRootMount(rootCtx, spec, conf, fds, mounts)
 	if err != nil {
-		return fmt.Errorf("failed to create root mount: %v", err)
+		return fmt.Errorf("creating root mount: %v", err)
 	}
 	mns, err := fs.NewMountNamespace(userCtx, rootInode)
 	if err != nil {
-		return fmt.Errorf("failed to create root mount namespace: %v", err)
+		return fmt.Errorf("creating root mount namespace: %v", err)
 	}
 	setMountNS(mns)
 
@@ -183,7 +183,7 @@ func createRootMount(ctx context.Context, spec *specs.Spec, conf *Config, fds *f
 	opts := p9MountOptions(fd, conf.FileAccess)
 	rootInode, err = p9FS.Mount(ctx, rootDevice, mf, strings.Join(opts, ","))
 	if err != nil {
-		return nil, fmt.Errorf("failed to generate root mount point: %v", err)
+		return nil, fmt.Errorf("creating root mount point: %v", err)
 	}
 
 	// We need to overlay the root on top of a ramfs with stub directories
@@ -192,7 +192,7 @@ func createRootMount(ctx context.Context, spec *specs.Spec, conf *Config, fds *f
 	submounts := append(subtargets("/", mounts), "/dev", "/sys", "/proc", "/tmp")
 	rootInode, err = addSubmountOverlay(ctx, rootInode, submounts)
 	if err != nil {
-		return nil, fmt.Errorf("error adding submount overlay: %v", err)
+		return nil, fmt.Errorf("adding submount overlay: %v", err)
 	}
 
 	if conf.Overlay && !spec.Root.Readonly {
@@ -204,7 +204,7 @@ func createRootMount(ctx context.Context, spec *specs.Spec, conf *Config, fds *f
 		}
 	}
 
-	log.Infof("Mounted %q to \"/\" type root", spec.Root.Path)
+	log.Infof("Mounted %q to %q type root", spec.Root.Path, "/")
 	return rootInode, nil
 }
 
@@ -222,7 +222,7 @@ func addOverlay(ctx context.Context, conf *Config, lower *fs.Inode, name string,
 	// Create overlay on top of mount dir.
 	upper, err := tmpFS.Mount(ctx, name+"-upper", lowerFlags, "")
 	if err != nil {
-		return nil, fmt.Errorf("failed to create tmpfs overlay: %v", err)
+		return nil, fmt.Errorf("creating tmpfs overlay: %v", err)
 	}
 	return fs.NewOverlayRoot(ctx, upper, lower, lowerFlags)
 }
@@ -311,7 +311,7 @@ func mountSubmount(ctx context.Context, conf *Config, mns *fs.MountNamespace, ro
 
 	inode, err := filesystem.Mount(ctx, mountDevice(m), mf, strings.Join(opts, ","))
 	if err != nil {
-		return fmt.Errorf("failed to create mount with source %q: %v", m.Source, err)
+		return fmt.Errorf("creating mount with source %q: %v", m.Source, err)
 	}
 
 	// If there are submounts, we need to overlay the mount on top of a
@@ -321,7 +321,7 @@ func mountSubmount(ctx context.Context, conf *Config, mns *fs.MountNamespace, ro
 		log.Infof("Adding submount overlay over %q", m.Destination)
 		inode, err = addSubmountOverlay(ctx, inode, submounts)
 		if err != nil {
-			return fmt.Errorf("error adding submount overlay: %v", err)
+			return fmt.Errorf("adding submount overlay: %v", err)
 		}
 	}
 
@@ -336,11 +336,11 @@ func mountSubmount(ctx context.Context, conf *Config, mns *fs.MountNamespace, ro
 	maxTraversals := uint(0)
 	dirent, err := mns.FindInode(ctx, root, root, m.Destination, &maxTraversals)
 	if err != nil {
-		return fmt.Errorf("failed to find mount destination %q: %v", m.Destination, err)
+		return fmt.Errorf("can't find mount destination %q: %v", m.Destination, err)
 	}
 	defer dirent.DecRef()
 	if err := mns.Mount(ctx, dirent, inode); err != nil {
-		return fmt.Errorf("failed to mount at destination %q: %v", m.Destination, err)
+		return fmt.Errorf("mount %q error: %v", m.Destination, err)
 	}
 
 	log.Infof("Mounted %q to %q type %s", m.Source, m.Destination, m.Type)
@@ -503,11 +503,11 @@ func addSubmountOverlay(ctx context.Context, inode *fs.Inode, submounts []string
 	msrc := fs.NewPseudoMountSource()
 	mountTree, err := ramfs.MakeDirectoryTree(ctx, msrc, submounts)
 	if err != nil {
-		return nil, fmt.Errorf("error creating mount tree: %v", err)
+		return nil, fmt.Errorf("creating mount tree: %v", err)
 	}
 	overlayInode, err := fs.NewOverlayRoot(ctx, inode, mountTree, fs.MountSourceFlags{})
 	if err != nil {
-		return nil, fmt.Errorf("failed to make mount overlay: %v", err)
+		return nil, fmt.Errorf("adding mount overlay: %v", err)
 	}
 	return overlayInode, err
 }
@@ -544,7 +544,7 @@ func setupContainerFS(procArgs *kernel.CreateProcessArgs, spec *specs.Spec, conf
 	// fd.
 	fdm, err := createFDMap(ctx, k, ls, console, stdioFDs)
 	if err != nil {
-		return fmt.Errorf("error importing fds: %v", err)
+		return fmt.Errorf("importing fds: %v", err)
 	}
 
 	// CreateProcess takes a reference on FDMap if successful. We
@@ -595,7 +595,7 @@ func setupContainerFS(procArgs *kernel.CreateProcessArgs, spec *specs.Spec, conf
 	fds := &fdDispenser{fds: goferFDs}
 	rootInode, err := createRootMount(rootCtx, spec, conf, fds, nil)
 	if err != nil {
-		return fmt.Errorf("error creating filesystem for container: %v", err)
+		return fmt.Errorf("creating filesystem for container: %v", err)
 	}
 
 	// Mount the container's root filesystem to the newly created mount point.
@@ -630,9 +630,10 @@ func setupContainerFS(procArgs *kernel.CreateProcessArgs, spec *specs.Spec, conf
 // executable matching the procArgs.Argv[0].
 func setExecutablePath(ctx context.Context, mns *fs.MountNamespace, procArgs *kernel.CreateProcessArgs) error {
 	paths := fs.GetPath(procArgs.Envv)
-	f, err := mns.ResolveExecutablePath(ctx, procArgs.WorkingDirectory, procArgs.Argv[0], paths)
+	exe := procArgs.Argv[0]
+	f, err := mns.ResolveExecutablePath(ctx, procArgs.WorkingDirectory, exe, paths)
 	if err != nil {
-		return err
+		return fmt.Errorf("searching for executable %q, cwd: %q, $PATH=%q: %v", exe, procArgs.WorkingDirectory, strings.Join(paths, ":"), err)
 	}
 	procArgs.Filename = f
 	return nil
@@ -666,7 +667,7 @@ func destroyContainerFS(ctx context.Context, cid string, k *kernel.Kernel) error
 		return nil
 	}
 	if err != nil {
-		return fmt.Errorf("error finding container root directory %q: %v", containerRoot, err)
+		return fmt.Errorf("finding container root directory %q: %v", containerRoot, err)
 	}
 	defer containerRootDirent.DecRef()
 
@@ -682,7 +683,7 @@ func destroyContainerFS(ctx context.Context, cid string, k *kernel.Kernel) error
 		log.Debugf("Unmounting container submount %q", root.BaseName())
 		m.FlushDirentRefs()
 		if err := mns.Unmount(ctx, root, true /* detach only */); err != nil && err != syserror.EINVAL {
-			return fmt.Errorf("error unmounting container submount %q: %v", root.BaseName(), err)
+			return fmt.Errorf("unmounting container submount %q: %v", root.BaseName(), err)
 		}
 	}
 
@@ -690,7 +691,7 @@ func destroyContainerFS(ctx context.Context, cid string, k *kernel.Kernel) error
 	log.Debugf("Unmounting container root %q", containerRoot)
 	containerRootDirent.Inode.MountSource.FlushDirentRefs()
 	if err := mns.Unmount(ctx, containerRootDirent, true /* detach only */); err != nil {
-		return fmt.Errorf("error unmounting container root mount %q: %v", containerRootDirent.BaseName(), err)
+		return fmt.Errorf("unmounting container root mount %q: %v", containerRootDirent.BaseName(), err)
 	}
 
 	// Get a reference to the parent directory and remove the root
@@ -698,12 +699,12 @@ func destroyContainerFS(ctx context.Context, cid string, k *kernel.Kernel) error
 	maxTraversals = 0
 	containersDirDirent, err := mns.FindInode(ctx, mnsRoot, nil, ChildContainersDir, &maxTraversals)
 	if err != nil {
-		return fmt.Errorf("error finding containers directory %q: %v", ChildContainersDir, err)
+		return fmt.Errorf("finding containers directory %q: %v", ChildContainersDir, err)
 	}
 	defer containersDirDirent.DecRef()
 	log.Debugf("Deleting container root %q", containerRoot)
 	if err := containersDirDirent.RemoveDirectory(ctx, mnsRoot, cid); err != nil {
-		return fmt.Errorf("error removing directory %q: %v", containerRoot, err)
+		return fmt.Errorf("removing directory %q: %v", containerRoot, err)
 	}
 
 	return nil
diff --git a/runsc/boot/loader.go b/runsc/boot/loader.go
index 71a2ab962..f3dc15f00 100644
--- a/runsc/boot/loader.go
+++ b/runsc/boot/loader.go
@@ -173,17 +173,17 @@ func New(args Args) (*Loader, error) {
 	// We initialize the rand package now to make sure /dev/urandom is pre-opened
 	// on kernels that do not support getrandom(2).
 	if err := rand.Init(); err != nil {
-		return nil, fmt.Errorf("error setting up rand: %v", err)
+		return nil, fmt.Errorf("setting up rand: %v", err)
 	}
 
 	if err := usage.Init(); err != nil {
-		return nil, fmt.Errorf("error setting up memory usage: %v", err)
+		return nil, fmt.Errorf("setting up memory usage: %v", err)
 	}
 
 	// Create kernel and platform.
 	p, err := createPlatform(args.Conf, args.DeviceFD)
 	if err != nil {
-		return nil, fmt.Errorf("error creating platform: %v", err)
+		return nil, fmt.Errorf("creating platform: %v", err)
 	}
 	k := &kernel.Kernel{
 		Platform: p,
@@ -194,18 +194,18 @@ func New(args Args) (*Loader, error) {
 	// Pass k as the platform since it is savable, unlike the actual platform.
 	vdso, err := loader.PrepareVDSO(k)
 	if err != nil {
-		return nil, fmt.Errorf("error creating vdso: %v", err)
+		return nil, fmt.Errorf("creating vdso: %v", err)
 	}
 
 	// Create timekeeper.
 	tk, err := kernel.NewTimekeeper(k, vdso.ParamPage.FileRange())
 	if err != nil {
-		return nil, fmt.Errorf("error creating timekeeper: %v", err)
+		return nil, fmt.Errorf("creating timekeeper: %v", err)
 	}
 	tk.SetClocks(time.NewCalibratedClocks())
 
 	if err := enableStrace(args.Conf); err != nil {
-		return nil, fmt.Errorf("failed to enable strace: %v", err)
+		return nil, fmt.Errorf("enabling strace: %v", err)
 	}
 
 	// Create an empty network stack because the network namespace may be empty at
@@ -214,13 +214,13 @@ func New(args Args) (*Loader, error) {
 	// Run().
 	networkStack, err := newEmptyNetworkStack(args.Conf, k)
 	if err != nil {
-		return nil, fmt.Errorf("failed to create network: %v", err)
+		return nil, fmt.Errorf("creating network: %v", err)
 	}
 
 	// Create capabilities.
 	caps, err := specutils.Capabilities(args.Spec.Process.Capabilities)
 	if err != nil {
-		return nil, fmt.Errorf("error creating capabilities: %v", err)
+		return nil, fmt.Errorf("converting capabilities: %v", err)
 	}
 
 	// Convert the spec's additional GIDs to KGIDs.
@@ -262,7 +262,7 @@ func New(args Args) (*Loader, error) {
 		RootIPCNamespace:            kernel.NewIPCNamespace(creds.UserNamespace),
 		RootAbstractSocketNamespace: kernel.NewAbstractSocketNamespace(),
 	}); err != nil {
-		return nil, fmt.Errorf("error initializing kernel: %v", err)
+		return nil, fmt.Errorf("initializing kernel: %v", err)
 	}
 
 	// Turn on packet logging if enabled.
@@ -279,11 +279,11 @@ func New(args Args) (*Loader, error) {
 
 	procArgs, err := newProcess(args.ID, args.Spec, creds, k)
 	if err != nil {
-		return nil, fmt.Errorf("failed to create init process for root container: %v", err)
+		return nil, fmt.Errorf("creating init process for root container: %v", err)
 	}
 
 	if err := initCompatLogs(args.UserLogFD); err != nil {
-		return nil, fmt.Errorf("init compat logs: %v", err)
+		return nil, fmt.Errorf("initializing compat logs: %v", err)
 	}
 
 	eid := execID{cid: args.ID}
@@ -303,7 +303,7 @@ func New(args Args) (*Loader, error) {
 	// We don't care about child signals; some platforms can generate a
 	// tremendous number of useless ones (I'm looking at you, ptrace).
 	if err := sighandling.IgnoreChildStop(); err != nil {
-		return nil, fmt.Errorf("failed to ignore child stop signals: %v", err)
+		return nil, fmt.Errorf("ignore child stop signals failed: %v", err)
 	}
 
 	// Handle signals by forwarding them to the root container process
@@ -353,7 +353,7 @@ func newProcess(id string, spec *specs.Spec, creds *auth.Credentials, k *kernel.
 	// Create initial limits.
 	ls, err := createLimitSet(spec)
 	if err != nil {
-		return kernel.CreateProcessArgs{}, fmt.Errorf("error creating limits: %v", err)
+		return kernel.CreateProcessArgs{}, fmt.Errorf("creating limits: %v", err)
 	}
 
 	// Create the process arguments.
@@ -441,7 +441,7 @@ func (l *Loader) run() error {
 			ControllerFD: l.ctrl.srv.FD(),
 		}
 		if err := filter.Install(opts); err != nil {
-			return fmt.Errorf("Failed to install seccomp filters: %v", err)
+			return fmt.Errorf("installing seccomp filters: %v", err)
 		}
 	}
 
@@ -465,13 +465,13 @@ func (l *Loader) run() error {
 		rootCtx := l.rootProcArgs.NewContext(l.k)
 		rootMns := l.k.RootMountNamespace()
 		if err := setExecutablePath(rootCtx, rootMns, &l.rootProcArgs); err != nil {
-			return fmt.Errorf("error setting executable path for %+v: %v", l.rootProcArgs, err)
+			return err
 		}
 
 		// Create the root container init task.
 		_, _, err := l.k.CreateProcess(l.rootProcArgs)
 		if err != nil {
-			return fmt.Errorf("failed to create init process: %v", err)
+			return fmt.Errorf("creating init process: %v", err)
 		}
 
 		// CreateProcess takes a reference on FDMap if successful.
@@ -521,7 +521,7 @@ func (l *Loader) startContainer(k *kernel.Kernel, spec *specs.Spec, conf *Config
 	// Create capabilities.
 	caps, err := specutils.Capabilities(spec.Process.Capabilities)
 	if err != nil {
-		return fmt.Errorf("error creating capabilities: %v", err)
+		return fmt.Errorf("creating capabilities: %v", err)
 	}
 
 	// Convert the spec's additional GIDs to KGIDs.
@@ -544,7 +544,7 @@ func (l *Loader) startContainer(k *kernel.Kernel, spec *specs.Spec, conf *Config
 
 	procArgs, err := newProcess(cid, spec, creds, l.k)
 	if err != nil {
-		return fmt.Errorf("failed to create new process: %v", err)
+		return fmt.Errorf("creating new process: %v", err)
 	}
 
 	// Can't take ownership away from os.File. dup them to get a new FDs.
@@ -570,20 +570,20 @@ func (l *Loader) startContainer(k *kernel.Kernel, spec *specs.Spec, conf *Config
 		procArgs.Limits,
 		k,
 		cid); err != nil {
-		return fmt.Errorf("failed to create new process: %v", err)
+		return fmt.Errorf("configuring container FS: %v", err)
 	}
 
 	// setFileSystemForProcess dup'd stdioFDs, so we can close them.
 	for i, fd := range stdioFDs {
 		if err := syscall.Close(fd); err != nil {
-			return fmt.Errorf("failed to close stdioFD #%d: %v", i, fd)
+			return fmt.Errorf("closing stdio FD #%d: %v", i, fd)
 		}
 	}
 
 	ctx := procArgs.NewContext(l.k)
 	mns := k.RootMountNamespace()
 	if err := setExecutablePath(ctx, mns, &procArgs); err != nil {
-		return fmt.Errorf("error setting executable path for %+v: %v", procArgs, err)
+		return fmt.Errorf("setting executable path for %+v: %v", procArgs, err)
 	}
 
 	l.mu.Lock()
@@ -596,7 +596,7 @@ func (l *Loader) startContainer(k *kernel.Kernel, spec *specs.Spec, conf *Config
 
 	tg, _, err := l.k.CreateProcess(procArgs)
 	if err != nil {
-		return fmt.Errorf("failed to create process in sentry: %v", err)
+		return fmt.Errorf("creating process: %v", err)
 	}
 	// CreateProcess takes a reference on FDMap if successful.
 	procArgs.FDMap.DecRef()
@@ -615,7 +615,7 @@ func (l *Loader) destroyContainer(cid string) error {
 	if _, _, err := l.threadGroupFromIDLocked(execID{cid: cid}); err == nil {
 		// If the container has started, kill and wait for all processes.
 		if err := l.signalAllProcesses(cid, int32(linux.SIGKILL)); err != nil {
-			return fmt.Errorf("failed to SIGKILL all container processes: %v", err)
+			return fmt.Errorf("sending SIGKILL to all container processes: %v", err)
 		}
 	}
 
@@ -628,7 +628,7 @@ func (l *Loader) destroyContainer(cid string) error {
 
 	ctx := l.rootProcArgs.NewContext(l.k)
 	if err := destroyContainerFS(ctx, cid, l.k); err != nil {
-		return fmt.Errorf("failed to destroy filesystem for container %q: %v", cid, err)
+		return fmt.Errorf("destroying filesystem for container %q: %v", cid, err)
 	}
 
 	// We made it!
@@ -715,11 +715,11 @@ func (l *Loader) waitPID(tgid kernel.ThreadID, cid string, clearStatus bool, wai
 	// In this case, find the process in the container's PID namespace.
 	initTG, _, err := l.threadGroupFromID(execID{cid: cid})
 	if err != nil {
-		return fmt.Errorf("failed to wait for PID %d: %v", tgid, err)
+		return fmt.Errorf("waiting for PID %d: %v", tgid, err)
 	}
 	tg := initTG.PIDNamespace().ThreadGroupWithID(tgid)
 	if tg == nil {
-		return fmt.Errorf("failed to wait for PID %d: no such process", tgid)
+		return fmt.Errorf("waiting for PID %d: no such process", tgid)
 	}
 	if tg.Leader().ContainerID() != cid {
 		return fmt.Errorf("process %d is part of a different container: %q", tgid, tg.Leader().ContainerID())
@@ -778,15 +778,21 @@ func newEmptyNetworkStack(conf *Config, clock tcpip.Clock) (inet.Stack, error) {
 // processes in the container, or to the foreground process group.
 func (l *Loader) signal(cid string, pid, signo int32, mode SignalDeliveryMode) error {
 	if pid < 0 {
-		return fmt.Errorf("failed to signal container %q PID %d: PID must be positive", cid, pid)
+		return fmt.Errorf("PID (%d) must be positive", pid)
 	}
 
 	switch mode {
 	case DeliverToProcess:
-		return l.signalProcess(cid, kernel.ThreadID(pid), signo)
+		if err := l.signalProcess(cid, kernel.ThreadID(pid), signo); err != nil {
+			return fmt.Errorf("signaling process in container %q PID %d: %v", cid, pid, err)
+		}
+		return nil
 
 	case DeliverToForegroundProcessGroup:
-		return l.signalForegrondProcessGroup(cid, kernel.ThreadID(pid), signo)
+		if err := l.signalForegrondProcessGroup(cid, kernel.ThreadID(pid), signo); err != nil {
+			return fmt.Errorf("signaling foreground process group in container %q PID %d: %v", cid, pid, err)
+		}
+		return nil
 
 	case DeliverToAllProcesses:
 		if pid != 0 {
@@ -795,12 +801,15 @@ func (l *Loader) signal(cid string, pid, signo int32, mode SignalDeliveryMode) e
 		// Check that the container has actually started before signaling it.
 		_, _, err := l.threadGroupFromID(execID{cid: cid})
 		if err != nil {
-			return fmt.Errorf("failed to signal container %q: %v", cid, err)
+			return err
 		}
-		return l.signalAllProcesses(cid, signo)
+		if err := l.signalAllProcesses(cid, signo); err != nil {
+			return fmt.Errorf("signaling all processes in container %q: %v", cid, err)
+		}
+		return nil
 
 	default:
-		panic(fmt.Sprintf("unknown signal signal delivery mode %v", mode))
+		panic(fmt.Sprintf("unknown signal delivery mode %v", mode))
 	}
 }
 
@@ -816,11 +825,11 @@ func (l *Loader) signalProcess(cid string, tgid kernel.ThreadID, signo int32) er
 	// signal it.
 	initTG, _, err := l.threadGroupFromID(execID{cid: cid})
 	if err != nil {
-		return fmt.Errorf("failed to signal container %q: %v", cid, err)
+		return fmt.Errorf("no thread group found: %v", err)
 	}
 	tg := initTG.PIDNamespace().ThreadGroupWithID(tgid)
 	if tg == nil {
-		return fmt.Errorf("failed to signal container %q PID %d: no such process", cid, tgid)
+		return fmt.Errorf("no such process with PID %d", tgid)
 	}
 	if tg.Leader().ContainerID() != cid {
 		return fmt.Errorf("process %d is part of a different container: %q", tgid, tg.Leader().ContainerID())
@@ -833,10 +842,10 @@ func (l *Loader) signalForegrondProcessGroup(cid string, tgid kernel.ThreadID, s
 	// and send the signal to it.
 	tg, tty, err := l.threadGroupFromID(execID{cid: cid, pid: tgid})
 	if err != nil {
-		return fmt.Errorf("failed to signal foreground process group for container %q PID %d: %v", cid, tgid, err)
+		return fmt.Errorf("no thread group found: %v", err)
 	}
 	if tty == nil {
-		return fmt.Errorf("failed to signal foreground process group in container %q PID %d: no TTY attached", cid, tgid)
+		return fmt.Errorf("no TTY attached")
 	}
 	pg := tty.ForegroundProcessGroup()
 	if pg == nil {
diff --git a/runsc/cgroup/cgroup.go b/runsc/cgroup/cgroup.go
index 65a0b6d7a..87f051e79 100644
--- a/runsc/cgroup/cgroup.go
+++ b/runsc/cgroup/cgroup.go
@@ -236,7 +236,7 @@ func (c *Cgroup) Uninstall() error {
 			}
 			return err
 		}, b); err != nil {
-			return fmt.Errorf("error removing cgroup path %q: %v", path, err)
+			return fmt.Errorf("removing cgroup path %q: %v", path, err)
 		}
 	}
 	return nil
diff --git a/runsc/cmd/boot.go b/runsc/cmd/boot.go
index fb1fd3e70..7f87b2623 100644
--- a/runsc/cmd/boot.go
+++ b/runsc/cmd/boot.go
@@ -154,7 +154,7 @@ func (b *Boot) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	defer specFile.Close()
 	spec, err := specutils.ReadSpecFromFile(b.bundleDir, specFile)
 	if err != nil {
-		Fatalf("error reading spec: %v", err)
+		Fatalf("reading spec: %v", err)
 	}
 	specutils.LogSpec(spec)
 
@@ -208,7 +208,7 @@ func (b *Boot) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	}
 	l, err := boot.New(bootArgs)
 	if err != nil {
-		Fatalf("error creating loader: %v", err)
+		Fatalf("creating loader: %v", err)
 	}
 
 	// Fatalf exits the process and doesn't run defers.
@@ -220,7 +220,7 @@ func (b *Boot) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	buf := make([]byte, 1)
 	if w, err := startSyncFile.Write(buf); err != nil || w != 1 {
 		l.Destroy()
-		Fatalf("Unable to write into the start-sync descriptor: %v", err)
+		Fatalf("unable to write into the start-sync descriptor: %v", err)
 	}
 	// Closes startSyncFile because 'l.Run()' only returns when the sandbox exits.
 	startSyncFile.Close()
@@ -231,7 +231,7 @@ func (b *Boot) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	// Run the application and wait for it to finish.
 	if err := l.Run(); err != nil {
 		l.Destroy()
-		Fatalf("error running sandbox: %v", err)
+		Fatalf("running sandbox: %v", err)
 	}
 
 	ws := l.WaitExit()
diff --git a/runsc/cmd/checkpoint.go b/runsc/cmd/checkpoint.go
index 4f4771da2..d8f748aa0 100644
--- a/runsc/cmd/checkpoint.go
+++ b/runsc/cmd/checkpoint.go
@@ -77,7 +77,7 @@ func (c *Checkpoint) Execute(_ context.Context, f *flag.FlagSet, args ...interfa
 
 	cont, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading container: %v", err)
+		Fatalf("loading container: %v", err)
 	}
 
 	if c.imagePath == "" {
@@ -85,7 +85,7 @@ func (c *Checkpoint) Execute(_ context.Context, f *flag.FlagSet, args ...interfa
 	}
 
 	if err := os.MkdirAll(c.imagePath, 0755); err != nil {
-		Fatalf("error making directories at path provided: %v", err)
+		Fatalf("making directories at path provided: %v", err)
 	}
 
 	fullImagePath := filepath.Join(c.imagePath, checkpointFileName)
@@ -115,12 +115,12 @@ func (c *Checkpoint) Execute(_ context.Context, f *flag.FlagSet, args ...interfa
 	// Restore into new container with same ID.
 	bundleDir := cont.BundleDir
 	if bundleDir == "" {
-		Fatalf("error setting bundleDir")
+		Fatalf("setting bundleDir")
 	}
 
 	spec, err := specutils.ReadSpec(bundleDir)
 	if err != nil {
-		Fatalf("error reading spec: %v", err)
+		Fatalf("reading spec: %v", err)
 	}
 
 	specutils.LogSpec(spec)
@@ -130,17 +130,17 @@ func (c *Checkpoint) Execute(_ context.Context, f *flag.FlagSet, args ...interfa
 	}
 
 	if err := cont.Destroy(); err != nil {
-		Fatalf("error destroying container: %v", err)
+		Fatalf("destroying container: %v", err)
 	}
 
 	cont, err = container.Create(id, spec, conf, bundleDir, "", "", "")
 	if err != nil {
-		Fatalf("error restoring container: %v", err)
+		Fatalf("restoring container: %v", err)
 	}
 	defer cont.Destroy()
 
 	if err := cont.Restore(spec, conf, fullImagePath); err != nil {
-		Fatalf("error starting container: %v", err)
+		Fatalf("starting container: %v", err)
 	}
 
 	ws, err := cont.Wait()
diff --git a/runsc/cmd/create.go b/runsc/cmd/create.go
index d187b8592..30c8fa283 100644
--- a/runsc/cmd/create.go
+++ b/runsc/cmd/create.go
@@ -89,7 +89,7 @@ func (c *Create) Execute(_ context.Context, f *flag.FlagSet, args ...interface{}
 	}
 	spec, err := specutils.ReadSpec(bundleDir)
 	if err != nil {
-		Fatalf("error reading spec: %v", err)
+		Fatalf("reading spec: %v", err)
 	}
 	specutils.LogSpec(spec)
 
@@ -97,7 +97,7 @@ func (c *Create) Execute(_ context.Context, f *flag.FlagSet, args ...interface{}
 	// container unless the metadata specifies that it should be run in an
 	// existing container.
 	if _, err := container.Create(id, spec, conf, bundleDir, c.consoleSocket, c.pidFile, c.userLog); err != nil {
-		Fatalf("error creating container: %v", err)
+		Fatalf("creating container: %v", err)
 	}
 	return subcommands.ExitSuccess
 }
diff --git a/runsc/cmd/debug.go b/runsc/cmd/debug.go
index de530c068..e10326754 100644
--- a/runsc/cmd/debug.go
+++ b/runsc/cmd/debug.go
@@ -68,7 +68,7 @@ func (d *Debug) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 		var err error
 		c, err = container.Load(conf.RootDir, f.Arg(0))
 		if err != nil {
-			Fatalf("error loading container %q: %v", f.Arg(0), err)
+			Fatalf("loading container %q: %v", f.Arg(0), err)
 		}
 	} else {
 		if f.NArg() != 0 {
@@ -78,12 +78,12 @@ func (d *Debug) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 		// Go over all sandboxes and find the one that matches PID.
 		ids, err := container.List(conf.RootDir)
 		if err != nil {
-			Fatalf("error listing containers: %v", err)
+			Fatalf("listing containers: %v", err)
 		}
 		for _, id := range ids {
 			candidate, err := container.Load(conf.RootDir, id)
 			if err != nil {
-				Fatalf("error loading container %q: %v", id, err)
+				Fatalf("loading container %q: %v", id, err)
 			}
 			if candidate.SandboxPid() == d.pid {
 				c = candidate
@@ -110,7 +110,7 @@ func (d *Debug) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 		log.Infof("Retrieving sandbox stacks")
 		stacks, err := c.Sandbox.Stacks()
 		if err != nil {
-			Fatalf("error retrieving stacks: %v", err)
+			Fatalf("retrieving stacks: %v", err)
 		}
 		log.Infof("     *** Stack dump ***\n%s", stacks)
 	}
diff --git a/runsc/cmd/delete.go b/runsc/cmd/delete.go
index 8c7c7a5cd..3206b267a 100644
--- a/runsc/cmd/delete.go
+++ b/runsc/cmd/delete.go
@@ -74,13 +74,13 @@ func (d *Delete) execute(ids []string, conf *boot.Config) error {
 				log.Warningf("couldn't find container %q: %v", id, err)
 				return nil
 			}
-			return fmt.Errorf("error loading container %q: %v", id, err)
+			return fmt.Errorf("loading container %q: %v", id, err)
 		}
 		if !d.force && c.Status != container.Created && c.Status != container.Stopped {
 			return fmt.Errorf("cannot delete container that is not stopped without --force flag")
 		}
 		if err := c.Destroy(); err != nil {
-			return fmt.Errorf("error destroying container: %v", err)
+			return fmt.Errorf("destroying container: %v", err)
 		}
 	}
 	return nil
diff --git a/runsc/cmd/events.go b/runsc/cmd/events.go
index a54856fb4..208d2f74b 100644
--- a/runsc/cmd/events.go
+++ b/runsc/cmd/events.go
@@ -76,7 +76,7 @@ func (evs *Events) Execute(ctx context.Context, f *flag.FlagSet, args ...interfa
 
 	c, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading sandbox: %v", err)
+		Fatalf("loading sandbox: %v", err)
 	}
 
 	// Repeatedly get stats from the container.
@@ -84,13 +84,13 @@ func (evs *Events) Execute(ctx context.Context, f *flag.FlagSet, args ...interfa
 		// Get the event and print it as JSON.
 		ev, err := c.Event()
 		if err != nil {
-			log.Warningf("error getting events for container: %v", err)
+			log.Warningf("Error getting events for container: %v", err)
 		}
 		// err must be preserved because it is used below when breaking
 		// out of the loop.
 		b, err := json.Marshal(ev)
 		if err != nil {
-			log.Warningf("error while marshalling event %v: %v", ev, err)
+			log.Warningf("Error while marshalling event %v: %v", ev, err)
 		} else {
 			os.Stdout.Write(b)
 		}
diff --git a/runsc/cmd/exec.go b/runsc/cmd/exec.go
index 548207222..13584d800 100644
--- a/runsc/cmd/exec.go
+++ b/runsc/cmd/exec.go
@@ -111,14 +111,14 @@ func (ex *Exec) SetFlags(f *flag.FlagSet) {
 func (ex *Exec) Execute(_ context.Context, f *flag.FlagSet, args ...interface{}) subcommands.ExitStatus {
 	e, id, err := ex.parseArgs(f)
 	if err != nil {
-		Fatalf("error parsing process spec: %v", err)
+		Fatalf("parsing process spec: %v", err)
 	}
 	conf := args[0].(*boot.Config)
 	waitStatus := args[1].(*syscall.WaitStatus)
 
 	c, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading sandbox: %v", err)
+		Fatalf("loading sandbox: %v", err)
 	}
 
 	// Replace empty settings with defaults from container.
@@ -128,13 +128,13 @@ func (ex *Exec) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	if e.Envv == nil {
 		e.Envv, err = resolveEnvs(c.Spec.Process.Env, ex.env)
 		if err != nil {
-			Fatalf("error getting environment variables: %v", err)
+			Fatalf("getting environment variables: %v", err)
 		}
 	}
 	if e.Capabilities == nil {
 		e.Capabilities, err = specutils.Capabilities(c.Spec.Process.Capabilities)
 		if err != nil {
-			Fatalf("error creating capabilities: %v", err)
+			Fatalf("creating capabilities: %v", err)
 		}
 	}
 
@@ -149,7 +149,7 @@ func (ex *Exec) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	// Start the new process and get it pid.
 	pid, err := c.Execute(e)
 	if err != nil {
-		Fatalf("error getting processes for container: %v", err)
+		Fatalf("getting processes for container: %v", err)
 	}
 
 	if e.StdioIsPty {
@@ -163,7 +163,7 @@ func (ex *Exec) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	if ex.internalPidFile != "" {
 		pidStr := []byte(strconv.Itoa(int(pid)))
 		if err := ioutil.WriteFile(ex.internalPidFile, pidStr, 0644); err != nil {
-			Fatalf("error writing internal pid file %q: %v", ex.internalPidFile, err)
+			Fatalf("writing internal pid file %q: %v", ex.internalPidFile, err)
 		}
 	}
 
@@ -172,14 +172,14 @@ func (ex *Exec) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	// returns.
 	if ex.pidFile != "" {
 		if err := ioutil.WriteFile(ex.pidFile, []byte(strconv.Itoa(os.Getpid())), 0644); err != nil {
-			Fatalf("error writing pid file: %v", err)
+			Fatalf("writing pid file: %v", err)
 		}
 	}
 
 	// Wait for the process to exit.
 	ws, err := c.WaitPID(pid, ex.clearStatus)
 	if err != nil {
-		Fatalf("error waiting on pid %d: %v", pid, err)
+		Fatalf("waiting on pid %d: %v", pid, err)
 	}
 	*waitStatus = ws
 	return subcommands.ExitSuccess
@@ -188,7 +188,7 @@ func (ex *Exec) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 func (ex *Exec) execAndWait(waitStatus *syscall.WaitStatus) subcommands.ExitStatus {
 	binPath, err := specutils.BinPath()
 	if err != nil {
-		Fatalf("error getting bin path: %v", err)
+		Fatalf("getting bin path: %v", err)
 	}
 	var args []string
 
@@ -199,7 +199,7 @@ func (ex *Exec) execAndWait(waitStatus *syscall.WaitStatus) subcommands.ExitStat
 	if pidFile == "" {
 		tmpDir, err := ioutil.TempDir("", "exec-pid-")
 		if err != nil {
-			Fatalf("error creating TempDir: %v", err)
+			Fatalf("creating TempDir: %v", err)
 		}
 		defer os.RemoveAll(tmpDir)
 		pidFile = filepath.Join(tmpDir, "pid")
@@ -232,7 +232,7 @@ func (ex *Exec) execAndWait(waitStatus *syscall.WaitStatus) subcommands.ExitStat
 		// socket.
 		tty, err := console.NewWithSocket(ex.consoleSocket)
 		if err != nil {
-			Fatalf("error setting up console with socket %q: %v", ex.consoleSocket, err)
+			Fatalf("setting up console with socket %q: %v", ex.consoleSocket, err)
 		}
 		defer tty.Close()
 
@@ -307,7 +307,7 @@ func (ex *Exec) argsFromCLI(argv []string) (*control.ExecArgs, error) {
 	for _, s := range ex.extraKGIDs {
 		kgid, err := strconv.Atoi(s)
 		if err != nil {
-			Fatalf("error parsing GID: %s, %v", s, err)
+			Fatalf("parsing GID: %s, %v", s, err)
 		}
 		extraKGIDs = append(extraKGIDs, auth.KGID(kgid))
 	}
diff --git a/runsc/cmd/gofer.go b/runsc/cmd/gofer.go
index 7276f3f26..43286a2e5 100644
--- a/runsc/cmd/gofer.go
+++ b/runsc/cmd/gofer.go
@@ -101,12 +101,12 @@ func (g *Gofer) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 
 	specFile, err := specutils.OpenCleanSpec(g.bundleDir)
 	if err != nil {
-		Fatalf("error opening spec: %v", err)
+		Fatalf("opening spec: %v", err)
 	}
 	spec, err := specutils.ReadSpecFromFile(g.bundleDir, specFile)
 	specFile.Close()
 	if err != nil {
-		Fatalf("error reading spec: %v", err)
+		Fatalf("reading spec: %v", err)
 	}
 	specutils.LogSpec(spec)
 
@@ -120,7 +120,7 @@ func (g *Gofer) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 		Fatalf("failed to chroot to %q: %v", root, err)
 	}
 	if err := syscall.Chdir("/"); err != nil {
-		Fatalf("failed to change working dir: %v", err)
+		Fatalf("changing working dir: %v", err)
 	}
 	log.Infof("Process chroot'd to %q", root)
 
@@ -131,7 +131,7 @@ func (g *Gofer) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 		PanicOnWrite: g.panicOnWrite,
 	})
 	if err != nil {
-		Fatalf("Error creating attach point: %v", err)
+		Fatalf("creating attach point: %v", err)
 	}
 	ats = append(ats, ap)
 	log.Infof("Serving %q mapped to %q on FD %d (ro: %t)", "/", root, g.ioFDs[0], spec.Root.Readonly)
@@ -145,12 +145,12 @@ func (g *Gofer) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 			}
 			ap, err := fsgofer.NewAttachPoint(m.Destination, cfg)
 			if err != nil {
-				Fatalf("Error creating attach point: %v", err)
+				Fatalf("creating attach point: %v", err)
 			}
 			ats = append(ats, ap)
 
 			if mountIdx >= len(g.ioFDs) {
-				Fatalf("No FD found for mount. Did you forget --io-fd? mount: %d, %v", len(g.ioFDs), m)
+				Fatalf("no FD found for mount. Did you forget --io-fd? mount: %d, %v", len(g.ioFDs), m)
 			}
 			log.Infof("Serving %q mapped on FD %d (ro: %t)", m.Destination, g.ioFDs[mountIdx], cfg.ROMount)
 			mountIdx++
@@ -161,7 +161,7 @@ func (g *Gofer) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	}
 
 	if err := filter.Install(); err != nil {
-		Fatalf("Failed to install seccomp filters: %v", err)
+		Fatalf("installing seccomp filters: %v", err)
 	}
 
 	runServers(ats, g.ioFDs)
@@ -176,7 +176,7 @@ func runServers(ats []p9.Attacher, ioFDs []int) {
 		go func(ioFD int, at p9.Attacher) {
 			socket, err := unet.NewSocket(ioFD)
 			if err != nil {
-				Fatalf("err creating server on FD %d: %v", ioFD, err)
+				Fatalf("creating server on FD %d: %v", ioFD, err)
 			}
 			s := p9.NewServer(at)
 			if err := s.Handle(socket); err != nil {
diff --git a/runsc/cmd/kill.go b/runsc/cmd/kill.go
index 7d86bb043..e67f82473 100644
--- a/runsc/cmd/kill.go
+++ b/runsc/cmd/kill.go
@@ -71,7 +71,7 @@ func (k *Kill) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 
 	c, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading container: %v", err)
+		Fatalf("loading container: %v", err)
 	}
 
 	// The OCI command-line spec says that the signal should be specified
diff --git a/runsc/cmd/list.go b/runsc/cmd/list.go
index acefcb2db..481066225 100644
--- a/runsc/cmd/list.go
+++ b/runsc/cmd/list.go
@@ -81,7 +81,7 @@ func (l *List) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	for _, id := range ids {
 		c, err := container.Load(conf.RootDir, id)
 		if err != nil {
-			Fatalf("error loading container %q: %v", id, err)
+			Fatalf("loading container %q: %v", id, err)
 		}
 		containers = append(containers, c)
 	}
@@ -108,7 +108,7 @@ func (l *List) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 			states = append(states, c.State())
 		}
 		if err := json.NewEncoder(os.Stdout).Encode(states); err != nil {
-			Fatalf("error marshaling container state: %v", err)
+			Fatalf("marshaling container state: %v", err)
 		}
 	default:
 		Fatalf("unknown list format %q", l.format)
diff --git a/runsc/cmd/path.go b/runsc/cmd/path.go
index baba937a8..1276f0dbd 100644
--- a/runsc/cmd/path.go
+++ b/runsc/cmd/path.go
@@ -22,7 +22,7 @@ import (
 func getwdOrDie() string {
 	wd, err := os.Getwd()
 	if err != nil {
-		Fatalf("error getting current working directory: %v", err)
+		Fatalf("getting current working directory: %v", err)
 	}
 	return wd
 }
diff --git a/runsc/cmd/pause.go b/runsc/cmd/pause.go
index ee608faba..2c93e5f3e 100644
--- a/runsc/cmd/pause.go
+++ b/runsc/cmd/pause.go
@@ -57,7 +57,7 @@ func (*Pause) Execute(_ context.Context, f *flag.FlagSet, args ...interface{}) s
 
 	cont, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading container: %v", err)
+		Fatalf("loading container: %v", err)
 	}
 
 	if err := cont.Pause(); err != nil {
diff --git a/runsc/cmd/ps.go b/runsc/cmd/ps.go
index fd76cf975..060d796f2 100644
--- a/runsc/cmd/ps.go
+++ b/runsc/cmd/ps.go
@@ -62,11 +62,11 @@ func (ps *PS) Execute(ctx context.Context, f *flag.FlagSet, args ...interface{})
 
 	c, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading sandbox: %v", err)
+		Fatalf("loading sandbox: %v", err)
 	}
 	pList, err := c.Processes()
 	if err != nil {
-		Fatalf("error getting processes for container: %v", err)
+		Fatalf("getting processes for container: %v", err)
 	}
 
 	switch ps.format {
@@ -75,11 +75,11 @@ func (ps *PS) Execute(ctx context.Context, f *flag.FlagSet, args ...interface{})
 	case "json":
 		o, err := control.PrintPIDsJSON(pList)
 		if err != nil {
-			Fatalf("error generating JSON: %v", err)
+			Fatalf("generating JSON: %v", err)
 		}
 		fmt.Println(o)
 	default:
-		Fatalf("Unsupported format: %s", ps.format)
+		Fatalf("unsupported format: %s", ps.format)
 	}
 
 	return subcommands.ExitSuccess
diff --git a/runsc/cmd/restore.go b/runsc/cmd/restore.go
index 64b302b0c..66b23c38e 100644
--- a/runsc/cmd/restore.go
+++ b/runsc/cmd/restore.go
@@ -84,7 +84,7 @@ func (r *Restore) Execute(_ context.Context, f *flag.FlagSet, args ...interface{
 	}
 	spec, err := specutils.ReadSpec(bundleDir)
 	if err != nil {
-		Fatalf("error reading spec: %v", err)
+		Fatalf("reading spec: %v", err)
 	}
 	specutils.LogSpec(spec)
 
@@ -96,15 +96,15 @@ func (r *Restore) Execute(_ context.Context, f *flag.FlagSet, args ...interface{
 
 	c, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading container: %v", err)
+		Fatalf("loading container: %v", err)
 	}
 	if err := c.Restore(spec, conf, restoreFile); err != nil {
-		Fatalf("error restoring container: %v", err)
+		Fatalf("restoring container: %v", err)
 	}
 
 	ws, err := c.Wait()
 	if err != nil {
-		Fatalf("error running container: %v", err)
+		Fatalf("running container: %v", err)
 	}
 	*waitStatus = ws
 
diff --git a/runsc/cmd/resume.go b/runsc/cmd/resume.go
index e684aeb5c..5551d1450 100644
--- a/runsc/cmd/resume.go
+++ b/runsc/cmd/resume.go
@@ -58,7 +58,7 @@ func (r *Resume) Execute(_ context.Context, f *flag.FlagSet, args ...interface{}
 
 	cont, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading container: %v", err)
+		Fatalf("loading container: %v", err)
 	}
 
 	if err := cont.Resume(); err != nil {
diff --git a/runsc/cmd/run.go b/runsc/cmd/run.go
index 9a574679f..be1c1b678 100644
--- a/runsc/cmd/run.go
+++ b/runsc/cmd/run.go
@@ -69,13 +69,13 @@ func (r *Run) Execute(_ context.Context, f *flag.FlagSet, args ...interface{}) s
 	}
 	spec, err := specutils.ReadSpec(bundleDir)
 	if err != nil {
-		Fatalf("error reading spec: %v", err)
+		Fatalf("reading spec: %v", err)
 	}
 	specutils.LogSpec(spec)
 
 	ws, err := container.Run(id, spec, conf, bundleDir, r.consoleSocket, r.pidFile, r.userLog)
 	if err != nil {
-		Fatalf("error running container: %v", err)
+		Fatalf("running container: %v", err)
 	}
 
 	*waitStatus = ws
diff --git a/runsc/cmd/spec.go b/runsc/cmd/spec.go
index ee306bfa6..063bd39c5 100644
--- a/runsc/cmd/spec.go
+++ b/runsc/cmd/spec.go
@@ -175,7 +175,7 @@ func (s *Spec) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	}
 
 	if err := ioutil.WriteFile(confPath, specTemplate, 0664); err != nil {
-		Fatalf("error writing to %q: %v", confPath, err)
+		Fatalf("writing to %q: %v", confPath, err)
 	}
 
 	return subcommands.ExitSuccess
diff --git a/runsc/cmd/start.go b/runsc/cmd/start.go
index 065efec06..9e2e0c11d 100644
--- a/runsc/cmd/start.go
+++ b/runsc/cmd/start.go
@@ -56,10 +56,10 @@ func (*Start) Execute(_ context.Context, f *flag.FlagSet, args ...interface{}) s
 
 	c, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading container: %v", err)
+		Fatalf("loading container: %v", err)
 	}
 	if err := c.Start(conf); err != nil {
-		Fatalf("error starting container: %v", err)
+		Fatalf("starting container: %v", err)
 	}
 	return subcommands.ExitSuccess
 }
diff --git a/runsc/cmd/state.go b/runsc/cmd/state.go
index 15e27b250..c3ef65ab5 100644
--- a/runsc/cmd/state.go
+++ b/runsc/cmd/state.go
@@ -59,7 +59,7 @@ func (*State) Execute(_ context.Context, f *flag.FlagSet, args ...interface{}) s
 
 	c, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading container: %v", err)
+		Fatalf("loading container: %v", err)
 	}
 	log.Debugf("Returning state for container %+v", c)
 
@@ -69,7 +69,7 @@ func (*State) Execute(_ context.Context, f *flag.FlagSet, args ...interface{}) s
 	// Write json-encoded state directly to stdout.
 	b, err := json.MarshalIndent(state, "", "  ")
 	if err != nil {
-		Fatalf("error marshaling container state: %v", err)
+		Fatalf("marshaling container state: %v", err)
 	}
 	os.Stdout.Write(b)
 	return subcommands.ExitSuccess
diff --git a/runsc/cmd/wait.go b/runsc/cmd/wait.go
index 1e1c1fe17..6498dd15c 100644
--- a/runsc/cmd/wait.go
+++ b/runsc/cmd/wait.go
@@ -66,7 +66,7 @@ func (wt *Wait) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	}
 	// You can't specify both -pid and -rootpid.
 	if wt.rootPID != unsetPID && wt.pid != unsetPID {
-		Fatalf("only up to one of -pid and -rootPid can be set")
+		Fatalf("only one of -pid and -rootPid can be set")
 	}
 
 	id := f.Arg(0)
@@ -74,7 +74,7 @@ func (wt *Wait) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 
 	c, err := container.Load(conf.RootDir, id)
 	if err != nil {
-		Fatalf("error loading container: %v", err)
+		Fatalf("loading container: %v", err)
 	}
 
 	var waitStatus syscall.WaitStatus
@@ -83,21 +83,21 @@ func (wt *Wait) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	case wt.rootPID == unsetPID && wt.pid == unsetPID:
 		ws, err := c.Wait()
 		if err != nil {
-			Fatalf("error waiting on container %q: %v", c.ID, err)
+			Fatalf("waiting on container %q: %v", c.ID, err)
 		}
 		waitStatus = ws
 	// Wait on a PID in the root PID namespace.
 	case wt.rootPID != unsetPID:
 		ws, err := c.WaitRootPID(int32(wt.rootPID), true /* clearStatus */)
 		if err != nil {
-			Fatalf("error waiting on PID in root PID namespace %d in container %q: %v", wt.rootPID, c.ID, err)
+			Fatalf("waiting on PID in root PID namespace %d in container %q: %v", wt.rootPID, c.ID, err)
 		}
 		waitStatus = ws
 	// Wait on a PID in the container's PID namespace.
 	case wt.pid != unsetPID:
 		ws, err := c.WaitPID(int32(wt.pid), true /* clearStatus */)
 		if err != nil {
-			Fatalf("error waiting on PID %d in container %q: %v", wt.pid, c.ID, err)
+			Fatalf("waiting on PID %d in container %q: %v", wt.pid, c.ID, err)
 		}
 		waitStatus = ws
 	}
@@ -107,7 +107,7 @@ func (wt *Wait) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	}
 	// Write json-encoded wait result directly to stdout.
 	if err := json.NewEncoder(os.Stdout).Encode(result); err != nil {
-		Fatalf("error marshaling wait result: %v", err)
+		Fatalf("marshaling wait result: %v", err)
 	}
 	return subcommands.ExitSuccess
 }
diff --git a/runsc/console/console.go b/runsc/console/console.go
index 9f4f9214d..2eb9a8807 100644
--- a/runsc/console/console.go
+++ b/runsc/console/console.go
@@ -30,7 +30,7 @@ func NewWithSocket(socketPath string) (*os.File, error) {
 	// Create a new pty master and slave.
 	ptyMaster, ptySlave, err := pty.Open()
 	if err != nil {
-		return nil, fmt.Errorf("error opening pty: %v", err)
+		return nil, fmt.Errorf("opening pty: %v", err)
 	}
 	defer ptyMaster.Close()
 
@@ -38,7 +38,7 @@ func NewWithSocket(socketPath string) (*os.File, error) {
 	conn, err := net.Dial("unix", socketPath)
 	if err != nil {
 		ptySlave.Close()
-		return nil, fmt.Errorf("error dial socket %q: %v", socketPath, err)
+		return nil, fmt.Errorf("dialing socket %q: %v", socketPath, err)
 	}
 	defer conn.Close()
 	uc, ok := conn.(*net.UnixConn)
@@ -49,7 +49,7 @@ func NewWithSocket(socketPath string) (*os.File, error) {
 	socket, err := uc.File()
 	if err != nil {
 		ptySlave.Close()
-		return nil, fmt.Errorf("error getting file for unix socket %v: %v", uc, err)
+		return nil, fmt.Errorf("getting file for unix socket %v: %v", uc, err)
 	}
 	defer socket.Close()
 
@@ -57,7 +57,7 @@ func NewWithSocket(socketPath string) (*os.File, error) {
 	msg := unix.UnixRights(int(ptyMaster.Fd()))
 	if err := unix.Sendmsg(int(socket.Fd()), []byte("pty-master"), msg, nil, 0); err != nil {
 		ptySlave.Close()
-		return nil, fmt.Errorf("error sending console over unix socket %q: %v", socketPath, err)
+		return nil, fmt.Errorf("sending console over unix socket %q: %v", socketPath, err)
 	}
 	return ptySlave, nil
 }
diff --git a/runsc/container/container.go b/runsc/container/container.go
index 544e7a250..2d4b85d9f 100644
--- a/runsc/container/container.go
+++ b/runsc/container/container.go
@@ -137,7 +137,7 @@ type Container struct {
 func Load(rootDir, id string) (*Container, error) {
 	log.Debugf("Load container %q %q", rootDir, id)
 	if err := validateID(id); err != nil {
-		return nil, fmt.Errorf("error validating id: %v", err)
+		return nil, fmt.Errorf("validating id: %v", err)
 	}
 
 	cRoot, err := findContainerRoot(rootDir, id)
@@ -162,11 +162,11 @@ func Load(rootDir, id string) (*Container, error) {
 			// Preserve error so that callers can distinguish 'not found' errors.
 			return nil, err
 		}
-		return nil, fmt.Errorf("error reading container metadata file %q: %v", metaFile, err)
+		return nil, fmt.Errorf("reading container metadata file %q: %v", metaFile, err)
 	}
 	var c Container
 	if err := json.Unmarshal(metaBytes, &c); err != nil {
-		return nil, fmt.Errorf("error unmarshaling container metadata from %q: %v", metaFile, err)
+		return nil, fmt.Errorf("unmarshaling container metadata from %q: %v", metaFile, err)
 	}
 
 	// If the status is "Running" or "Created", check that the sandbox
@@ -225,7 +225,7 @@ func List(rootDir string) ([]string, error) {
 	log.Debugf("List containers %q", rootDir)
 	fs, err := ioutil.ReadDir(rootDir)
 	if err != nil {
-		return nil, fmt.Errorf("ReadDir(%s) failed: %v", rootDir, err)
+		return nil, fmt.Errorf("reading dir %q: %v", rootDir, err)
 	}
 	var out []string
 	for _, f := range fs {
@@ -257,7 +257,7 @@ func Create(id string, spec *specs.Spec, conf *boot.Config, bundleDir, consoleSo
 	if _, err := os.Stat(filepath.Join(containerRoot, metadataFilename)); err == nil {
 		return nil, fmt.Errorf("container with id %q already exists", id)
 	} else if !os.IsNotExist(err) {
-		return nil, fmt.Errorf("error looking for existing container in %q: %v", containerRoot, err)
+		return nil, fmt.Errorf("looking for existing container in %q: %v", containerRoot, err)
 	}
 
 	c := &Container{
@@ -446,14 +446,14 @@ func Run(id string, spec *specs.Spec, conf *boot.Config, bundleDir, consoleSocke
 	log.Debugf("Run container %q in root dir: %s", id, conf.RootDir)
 	c, err := Create(id, spec, conf, bundleDir, consoleSocket, pidFile, userLog)
 	if err != nil {
-		return 0, fmt.Errorf("error creating container: %v", err)
+		return 0, fmt.Errorf("creating container: %v", err)
 	}
 	// Clean up partially created container if an error ocurrs.
 	// Any errors returned by Destroy() itself are ignored.
 	defer c.Destroy()
 
 	if err := c.Start(conf); err != nil {
-		return 0, fmt.Errorf("error starting container: %v", err)
+		return 0, fmt.Errorf("starting container: %v", err)
 	}
 	return c.Wait()
 }
@@ -595,7 +595,7 @@ func (c *Container) Pause() error {
 	}
 
 	if err := c.Sandbox.Pause(c.ID); err != nil {
-		return fmt.Errorf("error pausing container: %v", err)
+		return fmt.Errorf("pausing container: %v", err)
 	}
 	c.changeStatus(Paused)
 	return c.save()
@@ -615,7 +615,7 @@ func (c *Container) Resume() error {
 		return fmt.Errorf("cannot resume container %q in state %v", c.ID, c.Status)
 	}
 	if err := c.Sandbox.Resume(c.ID); err != nil {
-		return fmt.Errorf("error resuming container: %v", err)
+		return fmt.Errorf("resuming container: %v", err)
 	}
 	c.changeStatus(Running)
 	return c.save()
@@ -657,19 +657,19 @@ func (c *Container) Destroy() error {
 	var errs []string
 
 	if err := c.stop(); err != nil {
-		err = fmt.Errorf("error stopping container: %v", err)
+		err = fmt.Errorf("stopping container: %v", err)
 		log.Warningf("%v", err)
 		errs = append(errs, err.Error())
 	}
 
 	if err := destroyFS(c.Spec); err != nil {
-		err = fmt.Errorf("error destroying container fs: %v", err)
+		err = fmt.Errorf("destroying container fs: %v", err)
 		log.Warningf("%v", err)
 		errs = append(errs, err.Error())
 	}
 
 	if err := os.RemoveAll(c.Root); err != nil && !os.IsNotExist(err) {
-		err = fmt.Errorf("error deleting container root directory %q: %v", c.Root, err)
+		err = fmt.Errorf("deleting container root directory %q: %v", c.Root, err)
 		log.Warningf("%v", err)
 		errs = append(errs, err.Error())
 	}
@@ -702,10 +702,10 @@ func (c *Container) save() error {
 	metaFile := filepath.Join(c.Root, metadataFilename)
 	meta, err := json.Marshal(c)
 	if err != nil {
-		return fmt.Errorf("error marshaling container metadata: %v", err)
+		return fmt.Errorf("invalid container metadata: %v", err)
 	}
 	if err := ioutil.WriteFile(metaFile, meta, 0640); err != nil {
-		return fmt.Errorf("error writing container metadata: %v", err)
+		return fmt.Errorf("writing container metadata: %v", err)
 	}
 	return nil
 }
@@ -719,7 +719,7 @@ func (c *Container) stop() error {
 	if c.Sandbox != nil {
 		log.Debugf("Destroying container %q", c.ID)
 		if err := c.Sandbox.DestroyContainer(c.ID); err != nil {
-			return fmt.Errorf("error destroying container %q: %v", c.ID, err)
+			return fmt.Errorf("destroying container %q: %v", c.ID, err)
 		}
 		cgroup = c.Sandbox.Cgroup
 		// Only set sandbox to nil after it has been told to destroy the container.
@@ -917,12 +917,12 @@ func (c *Container) lock() (func() error, error) {
 // given container root directory.
 func lockContainerMetadata(containerRootDir string) (func() error, error) {
 	if err := os.MkdirAll(containerRootDir, 0711); err != nil {
-		return nil, fmt.Errorf("error creating container root directory %q: %v", containerRootDir, err)
+		return nil, fmt.Errorf("creating container root directory %q: %v", containerRootDir, err)
 	}
 	f := filepath.Join(containerRootDir, metadataLockFilename)
 	l := flock.NewFlock(f)
 	if err := l.Lock(); err != nil {
-		return nil, fmt.Errorf("error acquiring lock on container lock file %q: %v", f, err)
+		return nil, fmt.Errorf("acquiring lock on container lock file %q: %v", f, err)
 	}
 	return l.Unlock, nil
 }
diff --git a/runsc/container/fs.go b/runsc/container/fs.go
index 97195550f..998160487 100644
--- a/runsc/container/fs.go
+++ b/runsc/container/fs.go
@@ -94,14 +94,14 @@ func setupFS(spec *specs.Spec, conf *boot.Config, bundleDir string) ([]specs.Mou
 		flags |= syscall.MS_BIND
 		log.Infof("Mounting src: %q, dst: %q, flags: %#x", m.Source, dst, flags)
 		if err := specutils.Mount(m.Source, dst, m.Type, flags); err != nil {
-			return nil, fmt.Errorf("failed to mount %v: %v", m, err)
+			return nil, fmt.Errorf("mounting %v: %v", m, err)
 		}
 
 		// Make the mount a slave, so that for recursive bind mount, umount won't
 		// propagate to the source.
 		flags = syscall.MS_SLAVE | syscall.MS_REC
 		if err := syscall.Mount("", dst, "", uintptr(flags), ""); err != nil {
-			return nil, fmt.Errorf("failed to rslave mount dst: %q, flags: %#x, err: %v", dst, flags, err)
+			return nil, fmt.Errorf("mount rslave dst: %q, flags: %#x, err: %v", dst, flags, err)
 		}
 
 		cpy := m
@@ -146,7 +146,7 @@ func setupFS(spec *specs.Spec, conf *boot.Config, bundleDir string) ([]specs.Mou
 		flags := uintptr(syscall.MS_BIND | syscall.MS_REMOUNT | syscall.MS_RDONLY | syscall.MS_REC)
 		src := spec.Root.Path
 		if err := syscall.Mount(src, src, "bind", flags, ""); err != nil {
-			return nil, fmt.Errorf("failed to remount root as read-only with source: %q, target: %q, flags: %#x, err: %v", spec.Root.Path, spec.Root.Path, flags, err)
+			return nil, fmt.Errorf("remounting root as read-only with source: %q, target: %q, flags: %#x, err: %v", spec.Root.Path, spec.Root.Path, flags, err)
 		}
 	}
 	return rv, nil
diff --git a/runsc/sandbox/network.go b/runsc/sandbox/network.go
index 52fe8fc0f..8ec320d09 100644
--- a/runsc/sandbox/network.go
+++ b/runsc/sandbox/network.go
@@ -61,19 +61,19 @@ func setupNetwork(conn *urpc.Client, pid int, spec *specs.Spec, conf *boot.Confi
 	case boot.NetworkNone:
 		log.Infof("Network is disabled, create loopback interface only")
 		if err := createDefaultLoopbackInterface(conn); err != nil {
-			return fmt.Errorf("error creating default loopback interface: %v", err)
+			return fmt.Errorf("creating default loopback interface: %v", err)
 		}
 	case boot.NetworkSandbox:
 		// Build the path to the net namespace of the sandbox process.
 		// This is what we will copy.
 		nsPath := filepath.Join("/proc", strconv.Itoa(pid), "ns/net")
 		if err := createInterfacesAndRoutesFromNS(conn, nsPath); err != nil {
-			return fmt.Errorf("error creating interfaces from net namespace %q: %v", nsPath, err)
+			return fmt.Errorf("creating interfaces from net namespace %q: %v", nsPath, err)
 		}
 	case boot.NetworkHost:
 		// Nothing to do here.
 	default:
-		return fmt.Errorf("Invalid network type: %d", conf.Network)
+		return fmt.Errorf("invalid network type: %d", conf.Network)
 	}
 	return nil
 }
@@ -99,7 +99,7 @@ func createDefaultLoopbackInterface(conn *urpc.Client) error {
 	if err := conn.Call(boot.NetworkCreateLinksAndRoutes, &boot.CreateLinksAndRoutesArgs{
 		LoopbackLinks: []boot.LoopbackLink{link},
 	}, nil); err != nil {
-		return fmt.Errorf("error creating loopback link and routes: %v", err)
+		return fmt.Errorf("creating loopback link and routes: %v", err)
 	}
 	return nil
 }
@@ -112,7 +112,7 @@ func joinNetNS(nsPath string) (func(), error) {
 	})
 	if err != nil {
 		runtime.UnlockOSThread()
-		return nil, fmt.Errorf("error joining net namespace %q: %v", nsPath, err)
+		return nil, fmt.Errorf("joining net namespace %q: %v", nsPath, err)
 	}
 	return func() {
 		restoreNS()
@@ -147,7 +147,7 @@ func createInterfacesAndRoutesFromNS(conn *urpc.Client, nsPath string) error {
 	// Get all interfaces in the namespace.
 	ifaces, err := net.Interfaces()
 	if err != nil {
-		return fmt.Errorf("error querying interfaces: %v", err)
+		return fmt.Errorf("querying interfaces: %v", err)
 	}
 
 	if isRootNS(ifaces) {
@@ -164,14 +164,14 @@ func createInterfacesAndRoutesFromNS(conn *urpc.Client, nsPath string) error {
 
 		allAddrs, err := iface.Addrs()
 		if err != nil {
-			return fmt.Errorf("error fetching interface addresses for %q: %v", iface.Name, err)
+			return fmt.Errorf("fetching interface addresses for %q: %v", iface.Name, err)
 		}
 
 		// We build our own loopback devices.
 		if iface.Flags&net.FlagLoopback != 0 {
 			links, err := loopbackLinks(iface, allAddrs)
 			if err != nil {
-				return fmt.Errorf("error getting loopback routes and links for iface %q: %v", iface.Name, err)
+				return fmt.Errorf("getting loopback routes and links for iface %q: %v", iface.Name, err)
 			}
 			args.LoopbackLinks = append(args.LoopbackLinks, links...)
 			continue
@@ -218,7 +218,7 @@ func createInterfacesAndRoutesFromNS(conn *urpc.Client, nsPath string) error {
 		// will remove the routes as well.
 		routes, def, err := routesForIface(iface)
 		if err != nil {
-			return fmt.Errorf("error getting routes for interface %q: %v", iface.Name, err)
+			return fmt.Errorf("getting routes for interface %q: %v", iface.Name, err)
 		}
 		if def != nil {
 			if !args.DefaultGateway.Route.Empty() {
@@ -237,7 +237,7 @@ func createInterfacesAndRoutesFromNS(conn *urpc.Client, nsPath string) error {
 		// Get the link for the interface.
 		ifaceLink, err := netlink.LinkByName(iface.Name)
 		if err != nil {
-			return fmt.Errorf("error getting link for interface %q: %v", iface.Name, err)
+			return fmt.Errorf("getting link for interface %q: %v", iface.Name, err)
 		}
 
 		// Collect the addresses for the interface, enable forwarding,
@@ -247,7 +247,7 @@ func createInterfacesAndRoutesFromNS(conn *urpc.Client, nsPath string) error {
 
 			// Steal IP address from NIC.
 			if err := removeAddress(ifaceLink, addr.String()); err != nil {
-				return fmt.Errorf("error removing address %v from device %q: %v", iface.Name, addr, err)
+				return fmt.Errorf("removing address %v from device %q: %v", iface.Name, addr, err)
 			}
 		}
 
@@ -257,7 +257,7 @@ func createInterfacesAndRoutesFromNS(conn *urpc.Client, nsPath string) error {
 
 	log.Debugf("Setting up network, config: %+v", args)
 	if err := conn.Call(boot.NetworkCreateLinksAndRoutes, &args, nil); err != nil {
-		return fmt.Errorf("error creating links and routes: %v", err)
+		return fmt.Errorf("creating links and routes: %v", err)
 	}
 	return nil
 }
@@ -291,7 +291,7 @@ func routesForIface(iface net.Interface) ([]boot.Route, *boot.Route, error) {
 	}
 	rs, err := netlink.RouteList(link, netlink.FAMILY_ALL)
 	if err != nil {
-		return nil, nil, fmt.Errorf("error getting routes from %q: %v", iface.Name, err)
+		return nil, nil, fmt.Errorf("getting routes from %q: %v", iface.Name, err)
 	}
 
 	var def *boot.Route
diff --git a/runsc/sandbox/sandbox.go b/runsc/sandbox/sandbox.go
index df4c3c787..53cb464d2 100644
--- a/runsc/sandbox/sandbox.go
+++ b/runsc/sandbox/sandbox.go
@@ -100,7 +100,7 @@ func New(id string, spec *specs.Spec, conf *boot.Config, bundleDir, consoleSocke
 	// Wait until the sandbox has booted.
 	b := make([]byte, 1)
 	if l, err := clientSyncFile.Read(b); err != nil || l != 1 {
-		return nil, fmt.Errorf("error reading from the start-sync descriptor: %v", err)
+		return nil, fmt.Errorf("reading from the start-sync descriptor: %v", err)
 	}
 
 	c.Release()
@@ -133,13 +133,13 @@ func (s *Sandbox) StartRoot(spec *specs.Spec, conf *boot.Config) error {
 
 	// Configure the network.
 	if err := setupNetwork(conn, s.Pid, spec, conf); err != nil {
-		return fmt.Errorf("error setting up network: %v", err)
+		return fmt.Errorf("setting up network: %v", err)
 	}
 
 	// Send a message to the sandbox control server to start the root
 	// container.
 	if err := conn.Call(boot.RootContainerStart, &s.ID, nil); err != nil {
-		return fmt.Errorf("error starting root container %v: %v", spec.Process.Args, err)
+		return fmt.Errorf("starting root container: %v", err)
 	}
 
 	return nil
@@ -169,18 +169,18 @@ func (s *Sandbox) StartContainer(spec *specs.Spec, conf *boot.Config, cid string
 		FilePayload: urpc.FilePayload{Files: files},
 	}
 	if err := sandboxConn.Call(boot.ContainerStart, &args, nil); err != nil {
-		return fmt.Errorf("error starting non-root container %v: %v", spec.Process.Args, err)
+		return fmt.Errorf("starting non-root container %v: %v", spec.Process.Args, err)
 	}
 	return nil
 }
 
 // Restore sends the restore call for a container in the sandbox.
-func (s *Sandbox) Restore(cid string, spec *specs.Spec, conf *boot.Config, f string) error {
+func (s *Sandbox) Restore(cid string, spec *specs.Spec, conf *boot.Config, filename string) error {
 	log.Debugf("Restore sandbox %q", s.ID)
 
-	rf, err := os.Open(f)
+	rf, err := os.Open(filename)
 	if err != nil {
-		return fmt.Errorf("os.Open(%q) failed: %v", f, err)
+		return fmt.Errorf("opening restore file %q failed: %v", filename, err)
 	}
 	defer rf.Close()
 
@@ -207,12 +207,12 @@ func (s *Sandbox) Restore(cid string, spec *specs.Spec, conf *boot.Config, f str
 
 	// Configure the network.
 	if err := setupNetwork(conn, s.Pid, spec, conf); err != nil {
-		return fmt.Errorf("error setting up network: %v", err)
+		return fmt.Errorf("setting up network: %v", err)
 	}
 
 	// Restore the container and start the root container.
 	if err := conn.Call(boot.ContainerRestore, &opt, nil); err != nil {
-		return fmt.Errorf("error restoring container %q: %v", cid, err)
+		return fmt.Errorf("restoring container %q: %v", cid, err)
 	}
 
 	return nil
@@ -230,7 +230,7 @@ func (s *Sandbox) Processes(cid string) ([]*control.Process, error) {
 
 	var pl []*control.Process
 	if err := conn.Call(boot.ContainerProcesses, &cid, &pl); err != nil {
-		return nil, fmt.Errorf("error retrieving process data from sandbox: %v", err)
+		return nil, fmt.Errorf("retrieving process data from sandbox: %v", err)
 	}
 	return pl, nil
 }
@@ -248,7 +248,7 @@ func (s *Sandbox) Execute(args *control.ExecArgs) (int32, error) {
 	// Send a message to the sandbox control server to start the container.
 	var pid int32
 	if err := conn.Call(boot.ContainerExecuteAsync, args, &pid); err != nil {
-		return 0, fmt.Errorf("error executing command %q in sandbox: %v", args, err)
+		return 0, fmt.Errorf("executing command %q in sandbox: %v", args, err)
 	}
 	return pid, nil
 }
@@ -266,7 +266,7 @@ func (s *Sandbox) Event(cid string) (*boot.Event, error) {
 	// TODO: Pass in the container id (cid) here. The sandbox
 	// should return events only for that container.
 	if err := conn.Call(boot.ContainerEvent, nil, &e); err != nil {
-		return nil, fmt.Errorf("error retrieving event data from sandbox: %v", err)
+		return nil, fmt.Errorf("retrieving event data from sandbox: %v", err)
 	}
 	e.ID = cid
 	return &e, nil
@@ -282,7 +282,7 @@ func (s *Sandbox) sandboxConnect() (*urpc.Client, error) {
 }
 
 func (s *Sandbox) connError(err error) error {
-	return fmt.Errorf("error connecting to control server at PID %d: %v", s.Pid, err)
+	return fmt.Errorf("connecting to control server at PID %d: %v", s.Pid, err)
 }
 
 // createSandboxProcess starts the sandbox as a subprocess by running the "boot"
@@ -305,7 +305,7 @@ func (s *Sandbox) createSandboxProcess(spec *specs.Spec, conf *boot.Config, bund
 	if conf.LogFilename != "" {
 		logFile, err := os.OpenFile(conf.LogFilename, os.O_APPEND|os.O_CREATE|os.O_WRONLY, 0644)
 		if err != nil {
-			return fmt.Errorf("error opening log file %q: %v", conf.LogFilename, err)
+			return fmt.Errorf("opening log file %q: %v", conf.LogFilename, err)
 		}
 		defer logFile.Close()
 		cmd.ExtraFiles = append(cmd.ExtraFiles, logFile)
@@ -315,7 +315,7 @@ func (s *Sandbox) createSandboxProcess(spec *specs.Spec, conf *boot.Config, bund
 	if conf.DebugLog != "" {
 		debugLogFile, err := specutils.DebugLogFile(conf.DebugLog, "boot")
 		if err != nil {
-			return fmt.Errorf("error opening debug log file in %q: %v", conf.DebugLog, err)
+			return fmt.Errorf("opening debug log file in %q: %v", conf.DebugLog, err)
 		}
 		defer debugLogFile.Close()
 		cmd.ExtraFiles = append(cmd.ExtraFiles, debugLogFile)
@@ -333,7 +333,7 @@ func (s *Sandbox) createSandboxProcess(spec *specs.Spec, conf *boot.Config, bund
 	sockFD, err := server.CreateSocket(addr)
 	log.Infof("Creating sandbox process with addr: %s", addr[1:]) // skip "\00".
 	if err != nil {
-		return fmt.Errorf("error creating control server socket for sandbox %q: %v", s.ID, err)
+		return fmt.Errorf("creating control server socket for sandbox %q: %v", s.ID, err)
 	}
 	controllerFile := os.NewFile(uintptr(sockFD), "control_server_socket")
 	defer controllerFile.Close()
@@ -391,7 +391,7 @@ func (s *Sandbox) createSandboxProcess(spec *specs.Spec, conf *boot.Config, bund
 		// socket, and return the slave.
 		tty, err := console.NewWithSocket(consoleSocket)
 		if err != nil {
-			return fmt.Errorf("error setting up console with socket %q: %v", consoleSocket, err)
+			return fmt.Errorf("setting up console with socket %q: %v", consoleSocket, err)
 		}
 		defer tty.Close()
 
@@ -558,13 +558,13 @@ func (s *Sandbox) createSandboxProcess(spec *specs.Spec, conf *boot.Config, bund
 	if s.Cgroup != nil {
 		cpuNum, err := s.Cgroup.NumCPU()
 		if err != nil {
-			return fmt.Errorf("error getting cpu count from cgroups: %v", err)
+			return fmt.Errorf("getting cpu count from cgroups: %v", err)
 		}
 		cmd.Args = append(cmd.Args, "--cpu-num", strconv.Itoa(cpuNum))
 
 		mem, err := s.Cgroup.MemoryLimit()
 		if err != nil {
-			return fmt.Errorf("error getting memory limit from cgroups: %v", err)
+			return fmt.Errorf("getting memory limit from cgroups: %v", err)
 		}
 		// When memory limit is unset, a "large" number is returned. In that case,
 		// just stick with the default.
@@ -636,7 +636,7 @@ func (s *Sandbox) Wait(cid string) (syscall.WaitStatus, error) {
 		// "On Unix systems, FindProcess always succeeds and returns a
 		// Process for the given pid, regardless of whether the process
 		// exists."
-		return ws, fmt.Errorf("FindProcess(%d) failed: %v", s.Pid, err)
+		return ws, fmt.Errorf("Find process %d: %v", s.Pid, err)
 	}
 	ps, err := p.Wait()
 	if err != nil {
@@ -662,7 +662,7 @@ func (s *Sandbox) WaitPID(cid string, pid int32, clearStatus bool) (syscall.Wait
 		ClearStatus: clearStatus,
 	}
 	if err := conn.Call(boot.ContainerWaitPID, args, &ws); err != nil {
-		return ws, fmt.Errorf("error waiting on PID %d in sandbox %q: %v", pid, s.ID, err)
+		return ws, fmt.Errorf("waiting on PID %d in sandbox %q: %v", pid, s.ID, err)
 	}
 	return ws, nil
 }
@@ -680,10 +680,10 @@ func (s *Sandbox) destroy() error {
 	if s.Pid != 0 {
 		log.Debugf("Killing sandbox %q", s.ID)
 		if err := syscall.Kill(s.Pid, syscall.SIGKILL); err != nil && err != syscall.ESRCH {
-			return fmt.Errorf("error killing sandbox %q PID %q: %v", s.ID, s.Pid, err)
+			return fmt.Errorf("killing sandbox %q PID %q: %v", s.ID, s.Pid, err)
 		}
 		if err := s.waitForStopped(); err != nil {
-			return fmt.Errorf("error waiting sandbox %q stop: %v", s.ID, err)
+			return fmt.Errorf("waiting sandbox %q stop: %v", s.ID, err)
 		}
 	}
 
@@ -712,7 +712,7 @@ func (s *Sandbox) SignalContainer(cid string, sig syscall.Signal, all bool) erro
 		Mode:  mode,
 	}
 	if err := conn.Call(boot.ContainerSignal, &args, nil); err != nil {
-		return fmt.Errorf("err signaling container %q: %v", cid, err)
+		return fmt.Errorf("signaling container %q: %v", cid, err)
 	}
 	return nil
 }
@@ -741,7 +741,7 @@ func (s *Sandbox) SignalProcess(cid string, pid int32, sig syscall.Signal, fgPro
 		Mode:  mode,
 	}
 	if err := conn.Call(boot.ContainerSignal, &args, nil); err != nil {
-		return fmt.Errorf("err signaling container %q PID %d: %v", cid, pid, err)
+		return fmt.Errorf("signaling container %q PID %d: %v", cid, pid, err)
 	}
 	return nil
 }
@@ -763,7 +763,7 @@ func (s *Sandbox) Checkpoint(cid string, f *os.File) error {
 	}
 
 	if err := conn.Call(boot.ContainerCheckpoint, &opt, nil); err != nil {
-		return fmt.Errorf("err checkpointing container %q: %v", cid, err)
+		return fmt.Errorf("checkpointing container %q: %v", cid, err)
 	}
 	return nil
 }
@@ -778,7 +778,7 @@ func (s *Sandbox) Pause(cid string) error {
 	defer conn.Close()
 
 	if err := conn.Call(boot.ContainerPause, nil, nil); err != nil {
-		return fmt.Errorf("err pausing container %q: %v", cid, err)
+		return fmt.Errorf("pausing container %q: %v", cid, err)
 	}
 	return nil
 }
@@ -793,7 +793,7 @@ func (s *Sandbox) Resume(cid string) error {
 	defer conn.Close()
 
 	if err := conn.Call(boot.ContainerResume, nil, nil); err != nil {
-		return fmt.Errorf("err resuming container %q: %v", cid, err)
+		return fmt.Errorf("resuming container %q: %v", cid, err)
 	}
 	return nil
 }
@@ -821,7 +821,7 @@ func (s *Sandbox) Stacks() (string, error) {
 
 	var stacks string
 	if err := conn.Call(boot.SandboxStacks, nil, &stacks); err != nil {
-		return "", fmt.Errorf("err getting sandbox %q stacks: %v", s.ID, err)
+		return "", fmt.Errorf("getting sandbox %q stacks: %v", s.ID, err)
 	}
 	return stacks, nil
 }
@@ -846,7 +846,7 @@ func (s *Sandbox) DestroyContainer(cid string) error {
 	}
 	defer conn.Close()
 	if err := conn.Call(boot.ContainerDestroy, &cid, nil); err != nil {
-		return fmt.Errorf("error destroying container %q: %v", cid, err)
+		return fmt.Errorf("destroying container %q: %v", cid, err)
 	}
 	return nil
 }
@@ -889,7 +889,7 @@ func deviceFileForPlatform(p boot.PlatformType) (*os.File, error) {
 		return nil, nil
 	}
 	if err != nil {
-		return nil, fmt.Errorf("error opening device file for platform %q: %v", p, err)
+		return nil, fmt.Errorf("opening device file for platform %q: %v", p, err)
 	}
 	return f, err
 }