summaryrefslogtreecommitdiff
path: root/libpod/container_internal.go
blob: e22d36f996369926932a5ec3d0cde40d87a00e41 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
package libpod

import (
	"fmt"
	"io"
	"io/ioutil"
	"os"
	"path/filepath"
	"strings"
	"syscall"
	"time"

	"github.com/containers/storage"
	"github.com/containers/storage/pkg/archive"
	"github.com/containers/storage/pkg/chrootarchive"
	"github.com/docker/docker/pkg/mount"
	"github.com/docker/docker/pkg/namesgenerator"
	"github.com/docker/docker/pkg/stringid"
	spec "github.com/opencontainers/runtime-spec/specs-go"
	"github.com/opencontainers/runtime-tools/generate"
	"github.com/opencontainers/selinux/go-selinux/label"
	"github.com/pkg/errors"
	"github.com/sirupsen/logrus"
	"github.com/ulule/deepcopier"
	"golang.org/x/sys/unix"
)

const (
	// name of the directory holding the artifacts
	artifactsDir = "artifacts"
)

// rootFsSize gets the size of the container's root filesystem
// A container FS is split into two parts.  The first is the top layer, a
// mutable layer, and the rest is the RootFS: the set of immutable layers
// that make up the image on which the container is based.
func (c *Container) rootFsSize() (int64, error) {
	container, err := c.runtime.store.Container(c.ID())
	if err != nil {
		return 0, err
	}

	// Ignore the size of the top layer.   The top layer is a mutable RW layer
	// and is not considered a part of the rootfs
	rwLayer, err := c.runtime.store.Layer(container.LayerID)
	if err != nil {
		return 0, err
	}
	layer, err := c.runtime.store.Layer(rwLayer.Parent)
	if err != nil {
		return 0, err
	}

	size := int64(0)
	for layer.Parent != "" {
		layerSize, err := c.runtime.store.DiffSize(layer.Parent, layer.ID)
		if err != nil {
			return size, errors.Wrapf(err, "getting diffsize of layer %q and its parent %q", layer.ID, layer.Parent)
		}
		size += layerSize
		layer, err = c.runtime.store.Layer(layer.Parent)
		if err != nil {
			return 0, err
		}
	}
	// Get the size of the last layer.  Has to be outside of the loop
	// because the parent of the last layer is "", andlstore.Get("")
	// will return an error.
	layerSize, err := c.runtime.store.DiffSize(layer.Parent, layer.ID)
	return size + layerSize, err
}

// rwSize Gets the size of the mutable top layer of the container.
func (c *Container) rwSize() (int64, error) {
	container, err := c.runtime.store.Container(c.ID())
	if err != nil {
		return 0, err
	}

	// Get the size of the top layer by calculating the size of the diff
	// between the layer and its parent.  The top layer of a container is
	// the only RW layer, all others are immutable
	layer, err := c.runtime.store.Layer(container.LayerID)
	if err != nil {
		return 0, err
	}
	return c.runtime.store.DiffSize(layer.Parent, layer.ID)
}

// The path to the container's root filesystem - where the OCI spec will be
// placed, amongst other things
func (c *Container) bundlePath() string {
	return c.config.StaticDir
}

// Retrieves the path of the container's attach socket
func (c *Container) attachSocketPath() string {
	return filepath.Join(c.runtime.ociRuntime.socketsDir, c.ID(), "attach")
}

// Sync this container with on-disk state and runc status
// Should only be called with container lock held
// This function should suffice to ensure a container's state is accurate and
// it is valid for use.
func (c *Container) syncContainer() error {
	if err := c.runtime.state.UpdateContainer(c); err != nil {
		return err
	}
	// If runc knows about the container, update its status in runc
	// And then save back to disk
	if (c.state.State != ContainerStateUnknown) &&
		(c.state.State != ContainerStateConfigured) {
		oldState := c.state.State
		// TODO: optionally replace this with a stat for the exit file
		if err := c.runtime.ociRuntime.updateContainerStatus(c); err != nil {
			return err
		}
		// Only save back to DB if state changed
		if c.state.State != oldState {
			if err := c.save(); err != nil {
				return err
			}
		}
	}

	if !c.valid {
		return errors.Wrapf(ErrCtrRemoved, "container %s is not valid", c.ID())
	}

	return nil
}

// Make a new container
func newContainer(rspec *spec.Spec, lockDir string) (*Container, error) {
	if rspec == nil {
		return nil, errors.Wrapf(ErrInvalidArg, "must provide a valid runtime spec to create container")
	}

	ctr := new(Container)
	ctr.config = new(ContainerConfig)
	ctr.state = new(containerState)

	ctr.config.ID = stringid.GenerateNonCryptoID()
	ctr.config.Name = namesgenerator.GetRandomName(0)

	ctr.config.Spec = new(spec.Spec)
	deepcopier.Copy(rspec).To(ctr.config.Spec)
	ctr.config.CreatedTime = time.Now()

	ctr.config.ShmSize = DefaultShmSize
	ctr.config.CgroupParent = DefaultCgroupParent

	// Path our lock file will reside at
	lockPath := filepath.Join(lockDir, ctr.config.ID)
	// Grab a lockfile at the given path
	lock, err := storage.GetLockfile(lockPath)
	if err != nil {
		return nil, errors.Wrapf(err, "error creating lockfile for new container")
	}
	ctr.lock = lock

	return ctr, nil
}

// Create container root filesystem for use
func (c *Container) setupStorage() error {
	if !c.valid {
		return errors.Wrapf(ErrCtrRemoved, "container %s is not valid", c.ID())
	}

	if c.state.State != ContainerStateConfigured {
		return errors.Wrapf(ErrCtrStateInvalid, "container %s must be in Configured state to have storage set up", c.ID())
	}

	// Need both an image ID and image name, plus a bool telling us whether to use the image configuration
	if c.config.RootfsImageID == "" || c.config.RootfsImageName == "" {
		return errors.Wrapf(ErrInvalidArg, "must provide image ID and image name to use an image")
	}

	containerInfo, err := c.runtime.storageService.CreateContainerStorage(c.runtime.imageContext, c.config.RootfsImageName, c.config.RootfsImageID, c.config.Name, c.config.ID, c.config.MountLabel)
	if err != nil {
		return errors.Wrapf(err, "error creating container storage")
	}

	c.config.StaticDir = containerInfo.Dir
	c.state.RunDir = containerInfo.RunDir

	artifacts := filepath.Join(c.config.StaticDir, artifactsDir)
	if err := os.MkdirAll(artifacts, 0755); err != nil {
		return errors.Wrapf(err, "error creating artifacts directory %q", artifacts)
	}

	return nil
}

// Tear down a container's storage prior to removal
func (c *Container) teardownStorage() error {
	if !c.valid {
		return errors.Wrapf(ErrCtrRemoved, "container %s is not valid", c.ID())
	}

	if c.state.State == ContainerStateRunning || c.state.State == ContainerStatePaused {
		return errors.Wrapf(ErrCtrStateInvalid, "cannot remove storage for container %s as it is running or paused", c.ID())
	}

	artifacts := filepath.Join(c.config.StaticDir, artifactsDir)
	if err := os.RemoveAll(artifacts); err != nil {
		return errors.Wrapf(err, "error removing artifacts %q", artifacts)
	}

	if err := c.cleanupStorage(); err != nil {
		return errors.Wrapf(err, "failed to cleanup container %s storage", c.ID())
	}

	if err := c.runtime.storageService.DeleteContainer(c.ID()); err != nil {
		return errors.Wrapf(err, "error removing container %s root filesystem", c.ID())
	}

	return nil
}

// Refresh refreshes the container's state after a restart
func (c *Container) refresh() error {
	c.lock.Lock()
	defer c.lock.Unlock()

	if !c.valid {
		return errors.Wrapf(ErrCtrRemoved, "container %s is not valid - may have been removed", c.ID())
	}

	// We need to get the container's temporary directory from c/storage
	// It was lost in the reboot and must be recreated
	dir, err := c.runtime.storageService.GetRunDir(c.ID())
	if err != nil {
		return errors.Wrapf(err, "error retrieving temporary directory for container %s", c.ID())
	}
	c.state.RunDir = dir

	if err := c.runtime.state.SaveContainer(c); err != nil {
		return errors.Wrapf(err, "error refreshing state for container %s", c.ID())
	}

	return nil
}

func (c *Container) export(path string) error {
	mountPoint := c.state.Mountpoint
	if !c.state.Mounted {
		mount, err := c.runtime.store.Mount(c.ID(), c.config.MountLabel)
		if err != nil {
			return errors.Wrapf(err, "error mounting container %q", c.ID())
		}
		mountPoint = mount
		defer func() {
			if err := c.runtime.store.Unmount(c.ID()); err != nil {
				logrus.Errorf("error unmounting container %q: %v", c.ID(), err)
			}
		}()
	}

	input, err := archive.Tar(mountPoint, archive.Uncompressed)
	if err != nil {
		return errors.Wrapf(err, "error reading container directory %q", c.ID())
	}

	outFile, err := os.Create(path)
	if err != nil {
		return errors.Wrapf(err, "error creating file %q", path)
	}
	defer outFile.Close()

	_, err = io.Copy(outFile, input)
	return err
}

func (c *Container) addImageVolumes(g *generate.Generator) error {
	mountPoint := c.state.Mountpoint
	if !c.state.Mounted {
		return errors.Wrapf(ErrInternal, "container is not mounted")
	}

	imageStorage, err := c.runtime.getImage(c.config.RootfsImageID)
	if err != nil {
		return err
	}
	imageData, err := c.runtime.getImageInspectInfo(*imageStorage)
	if err != nil {
		return err
	}

	for k := range imageData.ContainerConfig.Volumes {
		mount := spec.Mount{
			Destination: k,
			Type:        "bind",
			Options:     []string{"rbind", "rw"},
		}
		if MountExists(g.Mounts(), k) {
			continue
		}
		volumePath := filepath.Join(c.config.StaticDir, "volumes", k)
		if _, err := os.Stat(volumePath); os.IsNotExist(err) {
			if err = os.MkdirAll(volumePath, 0755); err != nil {
				return errors.Wrapf(err, "error creating directory %q for volume %q in container %q", volumePath, k, c.ID)
			}
			if err = label.Relabel(volumePath, c.config.MountLabel, false); err != nil {
				return errors.Wrapf(err, "error relabeling directory %q for volume %q in container %q", volumePath, k, c.ID)
			}
			srcPath := filepath.Join(mountPoint, k)
			if err = chrootarchive.NewArchiver(nil).CopyWithTar(srcPath, volumePath); err != nil && !os.IsNotExist(err) {
				return errors.Wrapf(err, "error populating directory %q for volume %q in container %q using contents of %q", volumePath, k, c.ID, srcPath)
			}
			mount.Source = volumePath
		}
		g.AddMount(mount)
	}
	return nil
}

// Get path of artifact with a given name for this container
func (c *Container) getArtifactPath(name string) string {
	return filepath.Join(c.config.StaticDir, artifactsDir, name)
}

// Used with Wait() to determine if a container has exited
func (c *Container) isStopped() (bool, error) {
	if !c.locked {
		c.lock.Lock()
		defer c.lock.Unlock()
	}
	err := c.syncContainer()
	if err != nil {
		return true, err
	}
	return c.state.State == ContainerStateStopped, nil
}

// save container state to the database
func (c *Container) save() error {
	if err := c.runtime.state.SaveContainer(c); err != nil {
		return errors.Wrapf(err, "error saving container %s state", c.ID())
	}
	return nil
}

// Internal, non-locking function to stop container
func (c *Container) stop(timeout uint) error {
	logrus.Debugf("Stopping ctr %s with timeout %d", c.ID(), timeout)

	if c.state.State == ContainerStateConfigured ||
		c.state.State == ContainerStateUnknown ||
		c.state.State == ContainerStatePaused {
		return errors.Wrapf(ErrCtrStateInvalid, "can only stop created, running, or stopped containers")
	}

	if err := c.runtime.ociRuntime.stopContainer(c, timeout); err != nil {
		return err
	}

	// Sync the container's state to pick up return code
	if err := c.runtime.ociRuntime.updateContainerStatus(c); err != nil {
		return err
	}

	return c.cleanupStorage()
}

// mountStorage sets up the container's root filesystem
// It mounts the image and any other requested mounts
// TODO: Add ability to override mount label so we can use this for Mount() too
// TODO: Can we use this for export? Copying SHM into the export might not be
// good
func (c *Container) mountStorage() (err error) {
	// Container already mounted, nothing to do
	if c.state.Mounted {
		return nil
	}

	// TODO: generalize this mount code so it will mount every mount in ctr.config.Mounts

	mounted, err := mount.Mounted(c.config.ShmDir)
	if err != nil {
		return errors.Wrapf(err, "unable to determine if %q is mounted", c.config.ShmDir)
	}

	if !mounted {
		shmOptions := fmt.Sprintf("mode=1777,size=%d", c.config.ShmSize)
		if err := unix.Mount("shm", c.config.ShmDir, "tmpfs", unix.MS_NOEXEC|unix.MS_NOSUID|unix.MS_NODEV,
			label.FormatMountLabel(shmOptions, c.config.MountLabel)); err != nil {
			return errors.Wrapf(err, "failed to mount shm tmpfs %q", c.config.ShmDir)
		}
	}

	mountPoint, err := c.runtime.storageService.MountContainerImage(c.ID())
	if err != nil {
		return errors.Wrapf(err, "error mounting storage for container %s", c.ID())
	}
	c.state.Mounted = true
	c.state.Mountpoint = mountPoint

	logrus.Debugf("Created root filesystem for container %s at %s", c.ID(), c.state.Mountpoint)

	defer func() {
		if err != nil {
			if err2 := c.cleanupStorage(); err2 != nil {
				logrus.Errorf("Error unmounting storage for container %s: %v", c.ID(), err)
			}
		}
	}()

	return c.save()
}

// cleanupNetwork unmounts and cleans up the container's network
func (c *Container) cleanupNetwork() error {
	// Stop the container's network namespace (if it has one)
	if err := c.runtime.teardownNetNS(c); err != nil {
		logrus.Errorf("unable cleanup network for container %s: %q", c.ID(), err)
	}

	c.state.NetNS = nil
	c.state.SubnetMask = ""
	c.state.IPAddress = ""
	return c.save()
}

// cleanupStorage unmounts and cleans up the container's root filesystem
func (c *Container) cleanupStorage() error {
	if !c.state.Mounted {
		// Already unmounted, do nothing
		return nil
	}

	for _, mount := range c.config.Mounts {
		if err := unix.Unmount(mount, unix.MNT_DETACH); err != nil {
			if err != syscall.EINVAL {
				logrus.Warnf("container %s failed to unmount %s : %v", c.ID(), mount, err)
			}
		}
	}

	// Also unmount storage
	if err := c.runtime.storageService.UnmountContainerImage(c.ID()); err != nil {
		return errors.Wrapf(err, "error unmounting container %s root filesystem", c.ID())
	}

	c.state.Mountpoint = ""
	c.state.Mounted = false

	return c.save()
}

// WriteStringToRundir copies the provided file to the runtimedir
func (c *Container) WriteStringToRundir(destFile, output string) (string, error) {
	destFileName := filepath.Join(c.state.RunDir, destFile)
	f, err := os.Create(destFileName)
	if err != nil {
		return "", errors.Wrapf(err, "unable to create %s", destFileName)
	}

	defer f.Close()
	_, err = f.WriteString(output)
	if err != nil {
		return "", errors.Wrapf(err, "unable to write %s", destFileName)
	}
	// Relabel runDirResolv for the container
	if err := label.Relabel(destFileName, c.config.MountLabel, false); err != nil {
		return "", err
	}
	return destFileName, nil
}

type resolvConf struct {
	nameServers   []string
	searchDomains []string
	options       []string
}

// generateResolvConf generates a containers resolv.conf
func (c *Container) generateResolvConf() (string, error) {
	// Copy /etc/resolv.conf to the container's rundir
	resolvPath := "/etc/resolv.conf"

	// Check if the host system is using system resolve and if so
	// copy its resolv.conf
	if _, err := os.Stat("/run/systemd/resolve/resolv.conf"); err == nil {
		resolvPath = "/run/systemd/resolve/resolv.conf"
	}
	orig, err := ioutil.ReadFile(resolvPath)
	if err != nil {
		return "", errors.Wrapf(err, "unable to read %s", resolvPath)
	}
	if len(c.config.DNSServer) == 0 && len(c.config.DNSSearch) == 0 && len(c.config.DNSOption) == 0 {
		return c.WriteStringToRundir("resolv.conf", fmt.Sprintf("%s", orig))
	}

	// Read and organize the hosts /etc/resolv.conf
	resolv := createResolv(string(orig[:]))

	// Populate the resolv struct with user's dns search domains
	if len(c.config.DNSSearch) > 0 {
		resolv.searchDomains = nil
		// The . character means the user doesnt want any search domains in the container
		if !StringInSlice(".", c.config.DNSSearch) {
			resolv.searchDomains = append(resolv.searchDomains, c.Config().DNSSearch...)
		}
	}

	// Populate the resolv struct with user's dns servers
	if len(c.config.DNSServer) > 0 {
		resolv.nameServers = nil
		for _, i := range c.config.DNSServer {
			resolv.nameServers = append(resolv.nameServers, i.String())
		}
	}

	// Populate the resolve struct with the users dns options
	if len(c.config.DNSOption) > 0 {
		resolv.options = nil
		resolv.options = append(resolv.options, c.Config().DNSOption...)
	}
	return c.WriteStringToRundir("resolv.conf", resolv.ToString())
}

// createResolv creates a resolv struct from an input string
func createResolv(input string) resolvConf {
	var resolv resolvConf
	for _, line := range strings.Split(input, "\n") {
		if strings.HasPrefix(line, "search") {
			fields := strings.Fields(line)
			if len(fields) < 2 {
				logrus.Debugf("invalid resolv.conf line %s", line)
				continue
			}
			resolv.searchDomains = append(resolv.searchDomains, fields[1:]...)
		} else if strings.HasPrefix(line, "nameserver") {
			fields := strings.Fields(line)
			if len(fields) < 2 {
				logrus.Debugf("invalid resolv.conf line %s", line)
				continue
			}
			resolv.nameServers = append(resolv.nameServers, fields[1])
		} else if strings.HasPrefix(line, "options") {
			fields := strings.Fields(line)
			if len(fields) < 2 {
				logrus.Debugf("invalid resolv.conf line %s", line)
				continue
			}
			resolv.options = append(resolv.options, fields[1:]...)
		}
	}
	return resolv
}

//ToString returns a resolv struct in the form of a resolv.conf
func (r resolvConf) ToString() string {
	var result string
	// Populate the output string with search domains
	result += fmt.Sprintf("search %s\n", strings.Join(r.searchDomains, " "))
	// Populate the output string with name servers
	for _, i := range r.nameServers {
		result += fmt.Sprintf("nameserver %s\n", i)
	}
	// Populate the output string with dns options
	for _, i := range r.options {
		result += fmt.Sprintf("options %s\n", i)
	}
	return result
}

// generateHosts creates a containers hosts file
func (c *Container) generateHosts() (string, error) {
	orig, err := ioutil.ReadFile("/etc/hosts")
	if err != nil {
		return "", errors.Wrapf(err, "unable to read /etc/hosts")
	}
	hosts := string(orig)
	if len(c.config.HostAdd) > 0 {
		for _, host := range c.config.HostAdd {
			// the host format has already been verified at this point
			fields := strings.Split(host, ":")
			hosts += fmt.Sprintf("%s %s\n", fields[0], fields[1])
		}
	}
	return c.WriteStringToRundir("hosts", hosts)
}

// generateEtcHostname creates a containers /etc/hostname
func (c *Container) generateEtcHostname(hostname string) (string, error) {
	return c.WriteStringToRundir("hostname", hostname)
}