-
Notifications
You must be signed in to change notification settings - Fork 1.9k
/
executor_linux.go
885 lines (752 loc) · 23.2 KB
/
executor_linux.go
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
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
// +build linux
package executor
import (
"context"
"fmt"
"io"
"os"
"os/exec"
"path"
"path/filepath"
"strings"
"syscall"
"time"
"github.com/armon/circbuf"
"github.com/hashicorp/consul-template/signals"
hclog "github.com/hashicorp/go-hclog"
multierror "github.com/hashicorp/go-multierror"
"github.com/hashicorp/nomad/client/allocdir"
"github.com/hashicorp/nomad/client/stats"
cstructs "github.com/hashicorp/nomad/client/structs"
shelpers "github.com/hashicorp/nomad/helper/stats"
"github.com/hashicorp/nomad/helper/uuid"
"github.com/hashicorp/nomad/nomad/structs"
"github.com/hashicorp/nomad/plugins/drivers"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/cgroups"
cgroupFs "github.com/opencontainers/runc/libcontainer/cgroups/fs"
lconfigs "github.com/opencontainers/runc/libcontainer/configs"
ldevices "github.com/opencontainers/runc/libcontainer/devices"
lutils "github.com/opencontainers/runc/libcontainer/utils"
"github.com/syndtr/gocapability/capability"
"golang.org/x/sys/unix"
)
const (
defaultCgroupParent = "nomad"
)
var (
// ExecutorCgroupMeasuredMemStats is the list of memory stats captured by the executor
ExecutorCgroupMeasuredMemStats = []string{"RSS", "Cache", "Swap", "Usage", "Max Usage", "Kernel Usage", "Kernel Max Usage"}
// ExecutorCgroupMeasuredCpuStats is the list of CPU stats captures by the executor
ExecutorCgroupMeasuredCpuStats = []string{"System Mode", "User Mode", "Throttled Periods", "Throttled Time", "Percent"}
)
// LibcontainerExecutor implements an Executor with the runc/libcontainer api
type LibcontainerExecutor struct {
id string
command *ExecCommand
logger hclog.Logger
totalCpuStats *stats.CpuStats
userCpuStats *stats.CpuStats
systemCpuStats *stats.CpuStats
pidCollector *pidCollector
container libcontainer.Container
userProc *libcontainer.Process
userProcExited chan interface{}
exitState *ProcessState
}
func NewExecutorWithIsolation(logger hclog.Logger) Executor {
logger = logger.Named("isolated_executor")
if err := shelpers.Init(); err != nil {
logger.Error("unable to initialize stats", "error", err)
}
return &LibcontainerExecutor{
id: strings.Replace(uuid.Generate(), "-", "_", -1),
logger: logger,
totalCpuStats: stats.NewCpuStats(),
userCpuStats: stats.NewCpuStats(),
systemCpuStats: stats.NewCpuStats(),
pidCollector: newPidCollector(logger),
}
}
// Launch creates a new container in libcontainer and starts a new process with it
func (l *LibcontainerExecutor) Launch(command *ExecCommand) (*ProcessState, error) {
l.logger.Trace("preparing to launch command", "command", command.Cmd, "args", strings.Join(command.Args, " "))
if command.Resources == nil {
command.Resources = &drivers.Resources{
NomadResources: &structs.AllocatedTaskResources{},
}
}
l.command = command
// Move to the root cgroup until process is started
subsystems, err := cgroups.GetAllSubsystems()
if err != nil {
return nil, err
}
if err := JoinRootCgroup(subsystems); err != nil {
return nil, err
}
// create a new factory which will store the container state in the allocDir
factory, err := libcontainer.New(
path.Join(command.TaskDir, "../alloc/container"),
libcontainer.Cgroupfs,
// note that os.Args[0] refers to the executor shim typically
// and first args arguments is ignored now due
// until https://github.com/opencontainers/runc/pull/1888 is merged
libcontainer.InitArgs(os.Args[0], "libcontainer-shim"),
)
if err != nil {
return nil, fmt.Errorf("failed to create factory: %v", err)
}
// A container groups processes under the same isolation enforcement
containerCfg, err := newLibcontainerConfig(command)
if err != nil {
return nil, fmt.Errorf("failed to configure container(%s): %v", l.id, err)
}
container, err := factory.Create(l.id, containerCfg)
if err != nil {
return nil, fmt.Errorf("failed to create container(%s): %v", l.id, err)
}
l.container = container
// Look up the binary path and make it executable
absPath, err := lookupTaskBin(command)
if err != nil {
return nil, err
}
if err := makeExecutable(absPath); err != nil {
return nil, err
}
path := absPath
// Ensure that the path is contained in the chroot, and find it relative to the container
rel, err := filepath.Rel(command.TaskDir, path)
if err != nil {
return nil, fmt.Errorf("failed to determine relative path base=%q target=%q: %v", command.TaskDir, path, err)
}
// Turn relative-to-chroot path into absolute path to avoid
// libcontainer trying to resolve the binary using $PATH.
// Do *not* use filepath.Join as it will translate ".."s returned by
// filepath.Rel. Prepending "/" will cause the path to be rooted in the
// chroot which is the desired behavior.
path = "/" + rel
combined := append([]string{path}, command.Args...)
stdout, err := command.Stdout()
if err != nil {
return nil, err
}
stderr, err := command.Stderr()
if err != nil {
return nil, err
}
l.logger.Debug("launching", "command", command.Cmd, "args", strings.Join(command.Args, " "))
// the task process will be started by the container
process := &libcontainer.Process{
Args: combined,
Env: command.Env,
Stdout: stdout,
Stderr: stderr,
Init: true,
}
if command.User != "" {
process.User = command.User
}
l.userProc = process
l.totalCpuStats = stats.NewCpuStats()
l.userCpuStats = stats.NewCpuStats()
l.systemCpuStats = stats.NewCpuStats()
// Starts the task
if err := container.Run(process); err != nil {
container.Destroy()
return nil, err
}
pid, err := process.Pid()
if err != nil {
container.Destroy()
return nil, err
}
// Join process cgroups
containerState, err := container.State()
if err != nil {
l.logger.Error("error entering user process cgroups", "executor_pid", os.Getpid(), "error", err)
}
if err := cgroups.EnterPid(containerState.CgroupPaths, os.Getpid()); err != nil {
l.logger.Error("error entering user process cgroups", "executor_pid", os.Getpid(), "error", err)
}
// start a goroutine to wait on the process to complete, so Wait calls can
// be multiplexed
l.userProcExited = make(chan interface{})
go l.pidCollector.collectPids(l.userProcExited, l.getAllPids)
go l.wait()
return &ProcessState{
Pid: pid,
ExitCode: -1,
Time: time.Now(),
}, nil
}
func (l *LibcontainerExecutor) getAllPids() (map[int]*nomadPid, error) {
pids, err := l.container.Processes()
if err != nil {
return nil, err
}
nPids := make(map[int]*nomadPid)
for _, pid := range pids {
nPids[pid] = &nomadPid{
pid: pid,
cpuStatsTotal: stats.NewCpuStats(),
cpuStatsUser: stats.NewCpuStats(),
cpuStatsSys: stats.NewCpuStats(),
}
}
return nPids, nil
}
// Wait waits until a process has exited and returns it's exitcode and errors
func (l *LibcontainerExecutor) Wait(ctx context.Context) (*ProcessState, error) {
select {
case <-ctx.Done():
return nil, ctx.Err()
case <-l.userProcExited:
return l.exitState, nil
}
}
func (l *LibcontainerExecutor) wait() {
defer close(l.userProcExited)
ps, err := l.userProc.Wait()
if err != nil {
// If the process has exited before we called wait an error is returned
// the process state is embedded in the error
if exitErr, ok := err.(*exec.ExitError); ok {
ps = exitErr.ProcessState
} else {
l.logger.Error("failed to call wait on user process", "error", err)
l.exitState = &ProcessState{Pid: 0, ExitCode: 1, Time: time.Now()}
return
}
}
l.command.Close()
exitCode := 1
var signal int
if status, ok := ps.Sys().(syscall.WaitStatus); ok {
exitCode = status.ExitStatus()
if status.Signaled() {
const exitSignalBase = 128
signal = int(status.Signal())
exitCode = exitSignalBase + signal
}
}
l.exitState = &ProcessState{
Pid: ps.Pid(),
ExitCode: exitCode,
Signal: signal,
Time: time.Now(),
}
}
// Shutdown stops all processes started and cleans up any resources
// created (such as mountpoints, devices, etc).
func (l *LibcontainerExecutor) Shutdown(signal string, grace time.Duration) error {
if l.container == nil {
return nil
}
// move executor to root cgroup
subsystems, err := cgroups.GetAllSubsystems()
if err != nil {
return err
}
if err := JoinRootCgroup(subsystems); err != nil {
return err
}
status, err := l.container.Status()
if err != nil {
return err
}
defer l.container.Destroy()
if status == libcontainer.Stopped {
return nil
}
if grace > 0 {
if signal == "" {
signal = "SIGINT"
}
sig, ok := signals.SignalLookup[signal]
if !ok {
return fmt.Errorf("error unknown signal given for shutdown: %s", signal)
}
// Signal initial container processes only during graceful
// shutdown; hence `false` arg.
err = l.container.Signal(sig, false)
if err != nil {
return err
}
select {
case <-l.userProcExited:
return nil
case <-time.After(grace):
// Force kill all container processes after grace period,
// hence `true` argument.
if err := l.container.Signal(os.Kill, true); err != nil {
return err
}
}
} else {
if err := l.container.Signal(os.Kill, true); err != nil {
return err
}
}
select {
case <-l.userProcExited:
return nil
case <-time.After(time.Second * 15):
return fmt.Errorf("process failed to exit after 15 seconds")
}
}
// UpdateResources updates the resource isolation with new values to be enforced
func (l *LibcontainerExecutor) UpdateResources(resources *drivers.Resources) error {
return nil
}
// Version returns the api version of the executor
func (l *LibcontainerExecutor) Version() (*ExecutorVersion, error) {
return &ExecutorVersion{Version: ExecutorVersionLatest}, nil
}
// Stats returns the resource statistics for processes managed by the executor
func (l *LibcontainerExecutor) Stats(ctx context.Context, interval time.Duration) (<-chan *cstructs.TaskResourceUsage, error) {
ch := make(chan *cstructs.TaskResourceUsage)
go l.handleStats(ch, ctx, interval)
return ch, nil
}
func (l *LibcontainerExecutor) handleStats(ch chan *cstructs.TaskResourceUsage, ctx context.Context, interval time.Duration) {
defer close(ch)
timer := time.NewTimer(0)
for {
select {
case <-ctx.Done():
return
case <-timer.C:
timer.Reset(interval)
}
lstats, err := l.container.Stats()
if err != nil {
l.logger.Warn("error collecting stats", "error", err)
return
}
pidStats, err := l.pidCollector.pidStats()
if err != nil {
l.logger.Warn("error collecting stats", "error", err)
return
}
ts := time.Now()
stats := lstats.CgroupStats
// Memory Related Stats
swap := stats.MemoryStats.SwapUsage
maxUsage := stats.MemoryStats.Usage.MaxUsage
rss := stats.MemoryStats.Stats["rss"]
cache := stats.MemoryStats.Stats["cache"]
ms := &cstructs.MemoryStats{
RSS: rss,
Cache: cache,
Swap: swap.Usage,
Usage: stats.MemoryStats.Usage.Usage,
MaxUsage: maxUsage,
KernelUsage: stats.MemoryStats.KernelUsage.Usage,
KernelMaxUsage: stats.MemoryStats.KernelUsage.MaxUsage,
Measured: ExecutorCgroupMeasuredMemStats,
}
// CPU Related Stats
totalProcessCPUUsage := float64(stats.CpuStats.CpuUsage.TotalUsage)
userModeTime := float64(stats.CpuStats.CpuUsage.UsageInUsermode)
kernelModeTime := float64(stats.CpuStats.CpuUsage.UsageInKernelmode)
totalPercent := l.totalCpuStats.Percent(totalProcessCPUUsage)
cs := &cstructs.CpuStats{
SystemMode: l.systemCpuStats.Percent(kernelModeTime),
UserMode: l.userCpuStats.Percent(userModeTime),
Percent: totalPercent,
ThrottledPeriods: stats.CpuStats.ThrottlingData.ThrottledPeriods,
ThrottledTime: stats.CpuStats.ThrottlingData.ThrottledTime,
TotalTicks: l.systemCpuStats.TicksConsumed(totalPercent),
Measured: ExecutorCgroupMeasuredCpuStats,
}
taskResUsage := cstructs.TaskResourceUsage{
ResourceUsage: &cstructs.ResourceUsage{
MemoryStats: ms,
CpuStats: cs,
},
Timestamp: ts.UTC().UnixNano(),
Pids: pidStats,
}
select {
case <-ctx.Done():
return
case ch <- &taskResUsage:
}
}
}
// Signal sends a signal to the process managed by the executor
func (l *LibcontainerExecutor) Signal(s os.Signal) error {
return l.userProc.Signal(s)
}
// Exec starts an additional process inside the container
func (l *LibcontainerExecutor) Exec(deadline time.Time, cmd string, args []string) ([]byte, int, error) {
combined := append([]string{cmd}, args...)
// Capture output
buf, _ := circbuf.NewBuffer(int64(drivers.CheckBufSize))
process := &libcontainer.Process{
Args: combined,
Env: l.command.Env,
Stdout: buf,
Stderr: buf,
}
err := l.container.Run(process)
if err != nil {
return nil, 0, err
}
waitCh := make(chan *waitResult)
defer close(waitCh)
go l.handleExecWait(waitCh, process)
select {
case result := <-waitCh:
ps := result.ps
if result.err != nil {
if exitErr, ok := result.err.(*exec.ExitError); ok {
ps = exitErr.ProcessState
} else {
return nil, 0, result.err
}
}
var exitCode int
if status, ok := ps.Sys().(syscall.WaitStatus); ok {
exitCode = status.ExitStatus()
}
return buf.Bytes(), exitCode, nil
case <-time.After(time.Until(deadline)):
process.Signal(os.Kill)
return nil, 0, context.DeadlineExceeded
}
}
func (l *LibcontainerExecutor) newTerminalSocket() (pty func() (*os.File, error), tty *os.File, err error) {
parent, child, err := lutils.NewSockPair("socket")
if err != nil {
return nil, nil, fmt.Errorf("failed to create terminal: %v", err)
}
return func() (*os.File, error) { return lutils.RecvFd(parent) }, child, err
}
func (l *LibcontainerExecutor) ExecStreaming(ctx context.Context, cmd []string, tty bool,
stream drivers.ExecTaskStream) error {
// the task process will be started by the container
process := &libcontainer.Process{
Args: cmd,
Env: l.userProc.Env,
User: l.userProc.User,
Init: false,
Cwd: "/",
}
execHelper := &execHelper{
logger: l.logger,
newTerminal: l.newTerminalSocket,
setTTY: func(tty *os.File) error {
process.ConsoleSocket = tty
return nil
},
setIO: func(stdin io.Reader, stdout, stderr io.Writer) error {
process.Stdin = stdin
process.Stdout = stdout
process.Stderr = stderr
return nil
},
processStart: func() error { return l.container.Run(process) },
processWait: func() (*os.ProcessState, error) {
return process.Wait()
},
}
return execHelper.run(ctx, tty, stream)
}
type waitResult struct {
ps *os.ProcessState
err error
}
func (l *LibcontainerExecutor) handleExecWait(ch chan *waitResult, process *libcontainer.Process) {
ps, err := process.Wait()
ch <- &waitResult{ps, err}
}
func configureCapabilities(cfg *lconfigs.Config, command *ExecCommand) error {
// TODO: allow better control of these
// use capabilities list as prior to adopting libcontainer in 0.9
allCaps := supportedCaps()
// match capabilities used in Nomad 0.8
if command.User == "root" {
cfg.Capabilities = &lconfigs.Capabilities{
Bounding: allCaps,
Permitted: allCaps,
Effective: allCaps,
Ambient: nil,
Inheritable: nil,
}
} else {
cfg.Capabilities = &lconfigs.Capabilities{
Bounding: allCaps,
}
}
return nil
}
// supportedCaps returns a list of all supported capabilities in kernel
func supportedCaps() []string {
allCaps := []string{}
last := capability.CAP_LAST_CAP
// workaround for RHEL6 which has no /proc/sys/kernel/cap_last_cap
if last == capability.Cap(63) {
last = capability.CAP_BLOCK_SUSPEND
}
for _, cap := range capability.List() {
if cap > last {
continue
}
allCaps = append(allCaps, fmt.Sprintf("CAP_%s", strings.ToUpper(cap.String())))
}
return allCaps
}
// configureIsolation prepares the isolation primitives of the container.
// The process runs in a container configured with the following:
//
// * the task directory as the chroot
// * dedicated mount points namespace, but shares the PID, User, domain, network namespaces with host
// * small subset of devices (e.g. stdout/stderr/stdin, tty, shm, pts); default to using the same set of devices as Docker
// * some special filesystems: `/proc`, `/sys`. Some case is given to avoid exec escaping or setting malicious values through them.
func configureIsolation(cfg *lconfigs.Config, command *ExecCommand) error {
defaultMountFlags := syscall.MS_NOEXEC | syscall.MS_NOSUID | syscall.MS_NODEV
// set the new root directory for the container
cfg.Rootfs = command.TaskDir
// launch with mount namespace
cfg.Namespaces = lconfigs.Namespaces{
{Type: lconfigs.NEWNS},
}
// paths to mask using a bind mount to /dev/null to prevent reading
cfg.MaskPaths = []string{
"/proc/kcore",
"/sys/firmware",
}
// paths that should be remounted as readonly inside the container
cfg.ReadonlyPaths = []string{
"/proc/sys", "/proc/sysrq-trigger", "/proc/irq", "/proc/bus",
}
cfg.Devices = lconfigs.DefaultAutoCreatedDevices
if len(command.Devices) > 0 {
devs, err := cmdDevices(command.Devices)
if err != nil {
return err
}
cfg.Devices = append(cfg.Devices, devs...)
}
cfg.Mounts = []*lconfigs.Mount{
{
Source: "tmpfs",
Destination: "/dev",
Device: "tmpfs",
Flags: syscall.MS_NOSUID | syscall.MS_STRICTATIME,
Data: "mode=755",
},
{
Source: "proc",
Destination: "/proc",
Device: "proc",
Flags: defaultMountFlags,
},
{
Source: "devpts",
Destination: "/dev/pts",
Device: "devpts",
Flags: syscall.MS_NOSUID | syscall.MS_NOEXEC,
Data: "newinstance,ptmxmode=0666,mode=0620,gid=5",
},
{
Device: "tmpfs",
Source: "shm",
Destination: "/dev/shm",
Data: "mode=1777,size=65536k",
Flags: defaultMountFlags,
},
{
Source: "mqueue",
Destination: "/dev/mqueue",
Device: "mqueue",
Flags: defaultMountFlags,
},
{
Source: "sysfs",
Destination: "/sys",
Device: "sysfs",
Flags: defaultMountFlags | syscall.MS_RDONLY,
},
}
if len(command.Mounts) > 0 {
cfg.Mounts = append(cfg.Mounts, cmdMounts(command.Mounts)...)
}
return nil
}
func configureCgroups(cfg *lconfigs.Config, command *ExecCommand) error {
// If resources are not limited then manually create cgroups needed
if !command.ResourceLimits {
return configureBasicCgroups(cfg)
}
id := uuid.Generate()
cfg.Cgroups.Path = filepath.Join("/", defaultCgroupParent, id)
if command.Resources == nil || command.Resources.NomadResources == nil {
return nil
}
if mb := command.Resources.NomadResources.Memory.MemoryMB; mb > 0 {
// Total amount of memory allowed to consume
cfg.Cgroups.Resources.Memory = mb * 1024 * 1024
// Disable swap to avoid issues on the machine
var memSwappiness uint64
cfg.Cgroups.Resources.MemorySwappiness = &memSwappiness
}
cpuShares := command.Resources.NomadResources.Cpu.CpuShares
if cpuShares < 2 {
return fmt.Errorf("resources.Cpu.CpuShares must be equal to or greater than 2: %v", cpuShares)
}
// Set the relative CPU shares for this cgroup.
cfg.Cgroups.Resources.CpuShares = uint64(cpuShares)
return nil
}
func configureBasicCgroups(cfg *lconfigs.Config) error {
id := uuid.Generate()
// Manually create freezer cgroup
cfg.Cgroups.Paths = map[string]string{}
root, err := cgroups.FindCgroupMountpointDir()
if err != nil {
return err
}
if _, err := os.Stat(root); err != nil {
return err
}
freezer := cgroupFs.FreezerGroup{}
subsystem := freezer.Name()
path, err := cgroups.FindCgroupMountpoint("", subsystem)
if err != nil {
return fmt.Errorf("failed to find %s cgroup mountpoint: %v", subsystem, err)
}
// Sometimes subsystems can be mounted together as 'cpu,cpuacct'.
path = filepath.Join(root, filepath.Base(path), defaultCgroupParent, id)
if err = os.MkdirAll(path, 0755); err != nil {
return err
}
cfg.Cgroups.Paths[subsystem] = path
return nil
}
func newLibcontainerConfig(command *ExecCommand) (*lconfigs.Config, error) {
cfg := &lconfigs.Config{
Cgroups: &lconfigs.Cgroup{
Resources: &lconfigs.Resources{
AllowAllDevices: nil,
MemorySwappiness: nil,
AllowedDevices: lconfigs.DefaultAllowedDevices,
},
},
Version: "1.0.0",
}
if err := configureCapabilities(cfg, command); err != nil {
return nil, err
}
if err := configureIsolation(cfg, command); err != nil {
return nil, err
}
if err := configureCgroups(cfg, command); err != nil {
return nil, err
}
return cfg, nil
}
// JoinRootCgroup moves the current process to the cgroups of the init process
func JoinRootCgroup(subsystems []string) error {
mErrs := new(multierror.Error)
paths := map[string]string{}
for _, s := range subsystems {
mnt, _, err := cgroups.FindCgroupMountpointAndRoot("", s)
if err != nil {
multierror.Append(mErrs, fmt.Errorf("error getting cgroup path for subsystem: %s", s))
continue
}
paths[s] = mnt
}
err := cgroups.EnterPid(paths, os.Getpid())
if err != nil {
multierror.Append(mErrs, err)
}
return mErrs.ErrorOrNil()
}
// cmdDevices converts a list of driver.DeviceConfigs into excutor.Devices.
func cmdDevices(devices []*drivers.DeviceConfig) ([]*lconfigs.Device, error) {
if len(devices) == 0 {
return nil, nil
}
r := make([]*lconfigs.Device, len(devices))
for i, d := range devices {
ed, err := ldevices.DeviceFromPath(d.HostPath, d.Permissions)
if err != nil {
return nil, fmt.Errorf("failed to make device out for %s: %v", d.HostPath, err)
}
ed.Path = d.TaskPath
r[i] = ed
}
return r, nil
}
// cmdMounts converts a list of driver.MountConfigs into excutor.Mounts.
func cmdMounts(mounts []*drivers.MountConfig) []*lconfigs.Mount {
if len(mounts) == 0 {
return nil
}
r := make([]*lconfigs.Mount, len(mounts))
for i, m := range mounts {
flags := unix.MS_BIND
if m.Readonly {
flags |= unix.MS_RDONLY
}
r[i] = &lconfigs.Mount{
Source: m.HostPath,
Destination: m.TaskPath,
Device: "bind",
Flags: flags,
}
}
return r
}
// lookupTaskBin finds the file `bin` in taskDir/local, taskDir in that order, then performs
// a PATH search inside taskDir. It returns an absolute path. See also executor.lookupBin
func lookupTaskBin(command *ExecCommand) (string, error) {
taskDir := command.TaskDir
bin := command.Cmd
// Check in the local directory
localDir := filepath.Join(taskDir, allocdir.TaskLocal)
local := filepath.Join(localDir, bin)
if _, err := os.Stat(local); err == nil {
return local, nil
}
// Check at the root of the task's directory
root := filepath.Join(taskDir, bin)
if _, err := os.Stat(root); err == nil {
return root, nil
}
if strings.Contains(bin, "/") {
return "", fmt.Errorf("file %s not found under path %s", bin, taskDir)
}
// Find the PATH
path := "/usr/local/bin:/usr/bin:/bin"
for _, e := range command.Env {
if strings.HasPrefix("PATH=", e) {
path = e[5:]
}
}
return lookPathIn(path, taskDir, bin)
}
// lookPathIn looks for a file with PATH inside the directory root. Like exec.LookPath
func lookPathIn(path string, root string, bin string) (string, error) {
// exec.LookPath(file string)
for _, dir := range filepath.SplitList(path) {
if dir == "" {
// match unix shell behavior, empty path element == .
dir = "."
}
path := filepath.Join(root, dir, bin)
f, err := os.Stat(path)
if err != nil {
continue
}
if m := f.Mode(); !m.IsDir() {
return path, nil
}
}
return "", fmt.Errorf("file %s not found under path %s", bin, root)
}