aboutsummaryrefslogtreecommitdiff
path: root/pkg/specgen/generate/kube/kube.go
blob: 5862d3f1c2ce6b22b5883eab885b0e3e92fccb18 (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
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
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
package kube

import (
	"context"
	"encoding/json"
	"errors"
	"fmt"
	"math"
	"net"
	"regexp"
	"runtime"
	"strconv"
	"strings"
	"time"

	"github.com/containers/common/libimage"
	"github.com/containers/common/libnetwork/types"
	"github.com/containers/common/pkg/parse"
	"github.com/containers/common/pkg/secrets"
	cutil "github.com/containers/common/pkg/util"
	"github.com/containers/image/v5/manifest"
	"github.com/containers/podman/v4/libpod/define"
	ann "github.com/containers/podman/v4/pkg/annotations"
	"github.com/containers/podman/v4/pkg/domain/entities"
	v1 "github.com/containers/podman/v4/pkg/k8s.io/api/core/v1"
	"github.com/containers/podman/v4/pkg/k8s.io/apimachinery/pkg/api/resource"
	"github.com/containers/podman/v4/pkg/specgen"
	"github.com/containers/podman/v4/pkg/specgen/generate"
	"github.com/containers/podman/v4/pkg/util"
	"github.com/docker/docker/pkg/system"
	"github.com/docker/go-units"
	spec "github.com/opencontainers/runtime-spec/specs-go"
	"github.com/sirupsen/logrus"
)

func ToPodOpt(ctx context.Context, podName string, p entities.PodCreateOptions, podYAML *v1.PodTemplateSpec) (entities.PodCreateOptions, error) {
	p.Net = &entities.NetOptions{NoHosts: p.Net.NoHosts}

	p.Name = podName
	p.Labels = podYAML.ObjectMeta.Labels
	// Kube pods must share {ipc, net, uts} by default
	p.Share = append(p.Share, "ipc")
	p.Share = append(p.Share, "net")
	p.Share = append(p.Share, "uts")
	// TODO we only configure Process namespace. We also need to account for Host{IPC,Network,PID}
	// which is not currently possible with pod create
	if podYAML.Spec.ShareProcessNamespace != nil && *podYAML.Spec.ShareProcessNamespace {
		p.Share = append(p.Share, "pid")
	}
	p.Hostname = podYAML.Spec.Hostname
	if p.Hostname == "" {
		p.Hostname = podName
	}
	if podYAML.Spec.HostNetwork {
		p.Net.Network = specgen.Namespace{NSMode: "host"}
	}
	if podYAML.Spec.HostAliases != nil {
		if p.Net.NoHosts {
			return p, errors.New("HostAliases in yaml file will not work with --no-hosts")
		}
		hosts := make([]string, 0, len(podYAML.Spec.HostAliases))
		for _, hostAlias := range podYAML.Spec.HostAliases {
			for _, host := range hostAlias.Hostnames {
				hosts = append(hosts, host+":"+hostAlias.IP)
			}
		}
		p.Net.AddHosts = hosts
	}
	podPorts := getPodPorts(podYAML.Spec.Containers)
	p.Net.PublishPorts = podPorts

	if dnsConfig := podYAML.Spec.DNSConfig; dnsConfig != nil {
		// name servers
		if dnsServers := dnsConfig.Nameservers; len(dnsServers) > 0 {
			servers := make([]net.IP, 0)
			for _, server := range dnsServers {
				servers = append(servers, net.ParseIP(server))
			}
			p.Net.DNSServers = servers
		}
		// search domains
		if domains := dnsConfig.Searches; len(domains) > 0 {
			p.Net.DNSSearch = domains
		}
		// dns options
		if options := dnsConfig.Options; len(options) > 0 {
			dnsOptions := make([]string, 0, len(options))
			for _, opts := range options {
				d := opts.Name
				if opts.Value != nil {
					d += ":" + *opts.Value
				}
				dnsOptions = append(dnsOptions, d)
			}
			p.Net.DNSOptions = dnsOptions
		}
	}
	return p, nil
}

type CtrSpecGenOptions struct {
	// Annotations from the Pod
	Annotations map[string]string
	// Container as read from the pod yaml
	Container v1.Container
	// Image available to use (pulled or found local)
	Image *libimage.Image
	// Volumes for all containers
	Volumes map[string]*KubeVolume
	// PodID of the parent pod
	PodID string
	// PodName of the parent pod
	PodName string
	// PodInfraID as the infrastructure container id
	PodInfraID string
	// ConfigMaps the configuration maps for environment variables
	ConfigMaps []v1.ConfigMap
	// SeccompPaths for finding the seccomp profile path
	SeccompPaths *KubeSeccompPaths
	// RestartPolicy defines the restart policy of the container
	RestartPolicy string
	// NetNSIsHost tells the container to use the host netns
	NetNSIsHost bool
	// UserNSIsHost tells the container to use the host userns
	UserNSIsHost bool
	// SecretManager to access the secrets
	SecretsManager *secrets.SecretsManager
	// LogDriver which should be used for the container
	LogDriver string
	// LogOptions log options which should be used for the container
	LogOptions []string
	// Labels define key-value pairs of metadata
	Labels map[string]string
	//
	IsInfra bool
	// InitContainerType sets what type the init container is
	// Note: When playing a kube yaml, the inti container type will be set to "always" only
	InitContainerType string
	// PodSecurityContext is the security context specified for the pod
	PodSecurityContext *v1.PodSecurityContext
}

func ToSpecGen(ctx context.Context, opts *CtrSpecGenOptions) (*specgen.SpecGenerator, error) {
	s := specgen.NewSpecGenerator(opts.Container.Image, false)

	// pod name should be non-empty for Deployment objects to be able to create
	// multiple pods having containers with unique names
	if len(opts.PodName) < 1 {
		return nil, errors.New("got empty pod name on container creation when playing kube")
	}

	s.Name = fmt.Sprintf("%s-%s", opts.PodName, opts.Container.Name)

	s.Terminal = opts.Container.TTY

	s.Pod = opts.PodID

	s.LogConfiguration = &specgen.LogConfig{
		Driver: opts.LogDriver,
	}

	s.LogConfiguration.Options = make(map[string]string)
	for _, o := range opts.LogOptions {
		split := strings.SplitN(o, "=", 2)
		if len(split) < 2 {
			return nil, fmt.Errorf("invalid log option %q", o)
		}
		switch strings.ToLower(split[0]) {
		case "driver":
			s.LogConfiguration.Driver = split[1]
		case "path":
			s.LogConfiguration.Path = split[1]
		case "max-size":
			logSize, err := units.FromHumanSize(split[1])
			if err != nil {
				return nil, err
			}
			s.LogConfiguration.Size = logSize
		default:
			switch len(split[1]) {
			case 0:
				return nil, fmt.Errorf("invalid log option: %w", define.ErrInvalidArg)
			default:
				// tags for journald only
				if s.LogConfiguration.Driver == "" || s.LogConfiguration.Driver == define.JournaldLogging {
					s.LogConfiguration.Options[split[0]] = split[1]
				} else {
					logrus.Warnf("Can only set tags with journald log driver but driver is %q", s.LogConfiguration.Driver)
				}
			}
		}
	}

	s.InitContainerType = opts.InitContainerType

	setupSecurityContext(s, opts.Container.SecurityContext, opts.PodSecurityContext)
	err := setupLivenessProbe(s, opts.Container, opts.RestartPolicy)
	if err != nil {
		return nil, fmt.Errorf("failed to configure livenessProbe: %w", err)
	}

	// Since we prefix the container name with pod name to work-around the uniqueness requirement,
	// the seccomp profile should reference the actual container name from the YAML
	// but apply to the containers with the prefixed name
	s.SeccompProfilePath = opts.SeccompPaths.FindForContainer(opts.Container.Name)

	s.ResourceLimits = &spec.LinuxResources{}
	milliCPU, err := quantityToInt64(opts.Container.Resources.Limits.Cpu())
	if err != nil {
		return nil, fmt.Errorf("failed to set CPU quota: %w", err)
	}
	if milliCPU > 0 {
		period, quota := util.CoresToPeriodAndQuota(float64(milliCPU))
		s.ResourceLimits.CPU = &spec.LinuxCPU{
			Quota:  &quota,
			Period: &period,
		}
	}

	limit, err := quantityToInt64(opts.Container.Resources.Limits.Memory())
	if err != nil {
		return nil, fmt.Errorf("failed to set memory limit: %w", err)
	}

	memoryRes, err := quantityToInt64(opts.Container.Resources.Requests.Memory())
	if err != nil {
		return nil, fmt.Errorf("failed to set memory reservation: %w", err)
	}

	if limit > 0 || memoryRes > 0 {
		s.ResourceLimits.Memory = &spec.LinuxMemory{}
	}

	if limit > 0 {
		s.ResourceLimits.Memory.Limit = &limit
	}

	if memoryRes > 0 {
		s.ResourceLimits.Memory.Reservation = &memoryRes
	}

	// TODO: We don't understand why specgen does not take of this, but
	// integration tests clearly pointed out that it was required.
	imageData, err := opts.Image.Inspect(ctx, nil)
	if err != nil {
		return nil, err
	}
	s.WorkDir = "/"
	// Entrypoint/Command handling is based off of
	// https://kubernetes.io/docs/tasks/inject-data-application/define-command-argument-container/#notes
	if imageData != nil && imageData.Config != nil {
		if imageData.Config.WorkingDir != "" {
			s.WorkDir = imageData.Config.WorkingDir
		}
		if s.User == "" {
			s.User = imageData.Config.User
		}

		exposed, err := generate.GenExposedPorts(imageData.Config.ExposedPorts)
		if err != nil {
			return nil, err
		}

		for k, v := range s.Expose {
			exposed[k] = v
		}
		s.Expose = exposed
		// Pull entrypoint and cmd from image
		s.Entrypoint = imageData.Config.Entrypoint
		s.Command = imageData.Config.Cmd
		s.Labels = imageData.Config.Labels
		if len(imageData.Config.StopSignal) > 0 {
			stopSignal, err := util.ParseSignal(imageData.Config.StopSignal)
			if err != nil {
				return nil, err
			}
			s.StopSignal = &stopSignal
		}
	}
	// If only the yaml.Command is specified, set it as the entrypoint and drop the image Cmd
	if !opts.IsInfra && len(opts.Container.Command) != 0 {
		s.Entrypoint = opts.Container.Command
		s.Command = []string{}
	}
	// Only override the cmd field if yaml.Args is specified
	// Keep the image entrypoint, or the yaml.command if specified
	if !opts.IsInfra && len(opts.Container.Args) != 0 {
		s.Command = opts.Container.Args
	}

	// FIXME,
	// we are currently ignoring imageData.Config.ExposedPorts
	if !opts.IsInfra && opts.Container.WorkingDir != "" {
		s.WorkDir = opts.Container.WorkingDir
	}

	annotations := make(map[string]string)
	if opts.Annotations != nil {
		annotations = opts.Annotations
	}
	if opts.PodInfraID != "" {
		annotations[ann.SandboxID] = opts.PodInfraID
		annotations[ann.ContainerType] = ann.ContainerTypeContainer
	}
	s.Annotations = annotations

	// Environment Variables
	envs := map[string]string{}
	for _, env := range imageData.Config.Env {
		keyval := strings.SplitN(env, "=", 2)
		envs[keyval[0]] = keyval[1]
	}

	for _, env := range opts.Container.Env {
		value, err := envVarValue(env, opts)
		if err != nil {
			return nil, err
		}

		// Only set the env if the value is not nil
		if value != nil {
			envs[env.Name] = *value
		}
	}
	for _, envFrom := range opts.Container.EnvFrom {
		cmEnvs, err := envVarsFrom(envFrom, opts)
		if err != nil {
			return nil, err
		}

		for k, v := range cmEnvs {
			envs[k] = v
		}
	}
	s.Env = envs

	for _, volume := range opts.Container.VolumeMounts {
		volumeSource, exists := opts.Volumes[volume.Name]
		if !exists {
			return nil, fmt.Errorf("volume mount %s specified for container but not configured in volumes", volume.Name)
		}
		// Skip if the volume is optional. This means that a configmap for a configmap volume was not found but it was
		// optional so we can move on without throwing an error
		if exists && volumeSource.Optional {
			continue
		}

		dest, options, err := parseMountPath(volume.MountPath, volume.ReadOnly, volume.MountPropagation)
		if err != nil {
			return nil, err
		}

		volume.MountPath = dest
		switch volumeSource.Type {
		case KubeVolumeTypeBindMount:
			// If the container has bind mounts, we need to check if
			// a selinux mount option exists for it
			for k, v := range opts.Annotations {
				// Make sure the z/Z option is not already there (from editing the YAML)
				if k == define.BindMountPrefix {
					lastIndex := strings.LastIndex(v, ":")
					if v[:lastIndex] == volumeSource.Source && !cutil.StringInSlice("z", options) && !cutil.StringInSlice("Z", options) {
						options = append(options, v[lastIndex+1:])
					}
				}
			}
			mount := spec.Mount{
				Destination: volume.MountPath,
				Source:      volumeSource.Source,
				Type:        "bind",
				Options:     options,
			}
			s.Mounts = append(s.Mounts, mount)
		case KubeVolumeTypeNamed:
			namedVolume := specgen.NamedVolume{
				Dest:    volume.MountPath,
				Name:    volumeSource.Source,
				Options: options,
			}
			s.Volumes = append(s.Volumes, &namedVolume)
		case KubeVolumeTypeConfigMap:
			cmVolume := specgen.NamedVolume{
				Dest:    volume.MountPath,
				Name:    volumeSource.Source,
				Options: options,
			}
			s.Volumes = append(s.Volumes, &cmVolume)
		case KubeVolumeTypeCharDevice:
			// We are setting the path as hostPath:mountPath to comply with pkg/specgen/generate.DeviceFromPath.
			// The type is here just to improve readability as it is not taken into account when the actual device is created.
			device := spec.LinuxDevice{
				Path: fmt.Sprintf("%s:%s", volumeSource.Source, volume.MountPath),
				Type: "c",
			}
			s.Devices = append(s.Devices, device)
		case KubeVolumeTypeBlockDevice:
			// We are setting the path as hostPath:mountPath to comply with pkg/specgen/generate.DeviceFromPath.
			// The type is here just to improve readability as it is not taken into account when the actual device is created.
			device := spec.LinuxDevice{
				Path: fmt.Sprintf("%s:%s", volumeSource.Source, volume.MountPath),
				Type: "b",
			}
			s.Devices = append(s.Devices, device)
		case KubeVolumeTypeSecret:
			// in podman play kube we need to add these secrets as volumes rather than as
			// specgen.Secrets. Adding them as volumes allows for all key: value pairs to be mounted
			secretVolume := specgen.NamedVolume{
				Dest:    volume.MountPath,
				Name:    volumeSource.Source,
				Options: options,
			}
			s.Volumes = append(s.Volumes, &secretVolume)
		case KubeVolumeTypeEmptyDir:
			emptyDirVolume := specgen.NamedVolume{
				Dest:        volume.MountPath,
				Name:        volumeSource.Source,
				Options:     options,
				IsAnonymous: true,
			}
			s.Volumes = append(s.Volumes, &emptyDirVolume)
		default:
			return nil, errors.New("unsupported volume source type")
		}
	}

	s.RestartPolicy = opts.RestartPolicy

	if opts.NetNSIsHost {
		s.NetNS.NSMode = specgen.Host
	}
	if opts.UserNSIsHost {
		s.UserNS.NSMode = specgen.Host
	}

	// Add labels that come from kube
	if len(s.Labels) == 0 {
		// If there are no labels, let's use the map that comes
		// from kube
		s.Labels = opts.Labels
	} else {
		// If there are already labels in the map, append the ones
		// obtained from kube
		for k, v := range opts.Labels {
			s.Labels[k] = v
		}
	}

	return s, nil
}

func parseMountPath(mountPath string, readOnly bool, propagationMode *v1.MountPropagationMode) (string, []string, error) {
	options := []string{}
	splitVol := strings.Split(mountPath, ":")
	if len(splitVol) > 2 {
		return "", options, fmt.Errorf("%q incorrect volume format, should be ctr-dir[:option]", mountPath)
	}
	dest := splitVol[0]
	if len(splitVol) > 1 {
		options = strings.Split(splitVol[1], ",")
	}
	if err := parse.ValidateVolumeCtrDir(dest); err != nil {
		return "", options, fmt.Errorf("parsing MountPath: %w", err)
	}
	if readOnly {
		options = append(options, "ro")
	}
	opts, err := parse.ValidateVolumeOpts(options)
	if err != nil {
		return "", opts, fmt.Errorf("parsing MountOptions: %w", err)
	}
	if propagationMode != nil {
		switch *propagationMode {
		case v1.MountPropagationNone:
			opts = append(opts, "private")
		case v1.MountPropagationHostToContainer:
			opts = append(opts, "rslave")
		case v1.MountPropagationBidirectional:
			opts = append(opts, "rshared")
		default:
			return "", opts, fmt.Errorf("unknown propagation mode %q", *propagationMode)
		}
	}
	return dest, opts, nil
}

func setupLivenessProbe(s *specgen.SpecGenerator, containerYAML v1.Container, restartPolicy string) error {
	var err error
	if containerYAML.LivenessProbe == nil {
		return nil
	}
	emptyHandler := v1.Handler{}
	if containerYAML.LivenessProbe.Handler != emptyHandler {
		var commandString string
		failureCmd := "exit 1"
		probe := containerYAML.LivenessProbe
		probeHandler := probe.Handler

		// append `exit 1` to `cmd` so healthcheck can be marked as `unhealthy`.
		// append `kill 1` to `cmd` if appropriate restart policy is configured.
		if restartPolicy == "always" || restartPolicy == "onfailure" {
			// container will be restarted so we can kill init.
			failureCmd = "kill 1"
		}

		// configure healthcheck on the basis of Handler Actions.
		switch {
		case probeHandler.Exec != nil:
			execString := strings.Join(probeHandler.Exec.Command, " ")
			commandString = fmt.Sprintf("%s || %s", execString, failureCmd)
		case probeHandler.HTTPGet != nil:
			commandString = fmt.Sprintf("curl %s://%s:%d/%s  || %s", probeHandler.HTTPGet.Scheme, probeHandler.HTTPGet.Host, probeHandler.HTTPGet.Port.IntValue(), probeHandler.HTTPGet.Path, failureCmd)
		case probeHandler.TCPSocket != nil:
			commandString = fmt.Sprintf("nc -z -v %s %d || %s", probeHandler.TCPSocket.Host, probeHandler.TCPSocket.Port.IntValue(), failureCmd)
		}
		s.HealthConfig, err = makeHealthCheck(commandString, probe.PeriodSeconds, probe.FailureThreshold, probe.TimeoutSeconds, probe.InitialDelaySeconds)
		if err != nil {
			return err
		}
		return nil
	}
	return nil
}

func makeHealthCheck(inCmd string, interval int32, retries int32, timeout int32, startPeriod int32) (*manifest.Schema2HealthConfig, error) {
	// Every healthcheck requires a command
	if len(inCmd) == 0 {
		return nil, errors.New("must define a healthcheck command for all healthchecks")
	}

	// first try to parse option value as JSON array of strings...
	cmd := []string{}

	if inCmd == "none" {
		cmd = []string{define.HealthConfigTestNone}
	} else {
		err := json.Unmarshal([]byte(inCmd), &cmd)
		if err != nil {
			// ...otherwise pass it to "/bin/sh -c" inside the container
			cmd = []string{define.HealthConfigTestCmdShell}
			cmd = append(cmd, strings.Split(inCmd, " ")...)
		}
	}
	hc := manifest.Schema2HealthConfig{
		Test: cmd,
	}

	if interval < 1 {
		// kubernetes interval defaults to 10 sec and cannot be less than 1
		interval = 10
	}
	hc.Interval = (time.Duration(interval) * time.Second)
	if retries < 1 {
		// kubernetes retries defaults to 3
		retries = 3
	}
	hc.Retries = int(retries)
	if timeout < 1 {
		// kubernetes timeout defaults to 1
		timeout = 1
	}
	timeoutDuration := (time.Duration(timeout) * time.Second)
	if timeoutDuration < time.Duration(1) {
		return nil, errors.New("healthcheck-timeout must be at least 1 second")
	}
	hc.Timeout = timeoutDuration

	startPeriodDuration := (time.Duration(startPeriod) * time.Second)
	if startPeriodDuration < time.Duration(0) {
		return nil, errors.New("healthcheck-start-period must be 0 seconds or greater")
	}
	hc.StartPeriod = startPeriodDuration

	return &hc, nil
}

func setupSecurityContext(s *specgen.SpecGenerator, securityContext *v1.SecurityContext, podSecurityContext *v1.PodSecurityContext) {
	if securityContext == nil {
		securityContext = &v1.SecurityContext{}
	}
	if podSecurityContext == nil {
		podSecurityContext = &v1.PodSecurityContext{}
	}

	if securityContext.ReadOnlyRootFilesystem != nil {
		s.ReadOnlyFilesystem = *securityContext.ReadOnlyRootFilesystem
	}
	if securityContext.Privileged != nil {
		s.Privileged = *securityContext.Privileged
	}

	if securityContext.AllowPrivilegeEscalation != nil {
		s.NoNewPrivileges = !*securityContext.AllowPrivilegeEscalation
	}

	seopt := securityContext.SELinuxOptions
	if seopt == nil {
		seopt = podSecurityContext.SELinuxOptions
	}
	if seopt != nil {
		if seopt.User != "" {
			s.SelinuxOpts = append(s.SelinuxOpts, fmt.Sprintf("user:%s", seopt.User))
		}
		if seopt.Role != "" {
			s.SelinuxOpts = append(s.SelinuxOpts, fmt.Sprintf("role:%s", seopt.Role))
		}
		if seopt.Type != "" {
			s.SelinuxOpts = append(s.SelinuxOpts, fmt.Sprintf("type:%s", seopt.Type))
		}
		if seopt.Level != "" {
			s.SelinuxOpts = append(s.SelinuxOpts, fmt.Sprintf("level:%s", seopt.Level))
		}
	}
	if caps := securityContext.Capabilities; caps != nil {
		for _, capability := range caps.Add {
			s.CapAdd = append(s.CapAdd, string(capability))
		}
		for _, capability := range caps.Drop {
			s.CapDrop = append(s.CapDrop, string(capability))
		}
	}
	runAsUser := securityContext.RunAsUser
	if runAsUser == nil {
		runAsUser = podSecurityContext.RunAsUser
	}
	if runAsUser != nil {
		s.User = fmt.Sprintf("%d", *runAsUser)
	}

	runAsGroup := securityContext.RunAsGroup
	if runAsGroup == nil {
		runAsGroup = podSecurityContext.RunAsGroup
	}
	if runAsGroup != nil {
		if s.User == "" {
			s.User = "0"
		}
		s.User = fmt.Sprintf("%s:%d", s.User, *runAsGroup)
	}
	for _, group := range podSecurityContext.SupplementalGroups {
		s.Groups = append(s.Groups, fmt.Sprintf("%d", group))
	}
}

func quantityToInt64(quantity *resource.Quantity) (int64, error) {
	if i, ok := quantity.AsInt64(); ok {
		return i, nil
	}

	if i, ok := quantity.AsDec().Unscaled(); ok {
		return i, nil
	}

	return 0, fmt.Errorf("quantity cannot be represented as int64: %v", quantity)
}

// read a k8s secret in JSON format from the secret manager
func k8sSecretFromSecretManager(name string, secretsManager *secrets.SecretsManager) (map[string][]byte, error) {
	_, jsonSecret, err := secretsManager.LookupSecretData(name)
	if err != nil {
		return nil, err
	}

	var secrets map[string][]byte
	if err := json.Unmarshal(jsonSecret, &secrets); err != nil {
		return nil, fmt.Errorf("secret %v is not valid JSON: %v", name, err)
	}
	return secrets, nil
}

// envVarsFrom returns all key-value pairs as env vars from a configMap or secret that matches the envFrom setting of a container
func envVarsFrom(envFrom v1.EnvFromSource, opts *CtrSpecGenOptions) (map[string]string, error) {
	envs := map[string]string{}

	if envFrom.ConfigMapRef != nil {
		cmRef := envFrom.ConfigMapRef
		err := fmt.Errorf("configmap %v not found", cmRef.Name)

		for _, c := range opts.ConfigMaps {
			if cmRef.Name == c.Name {
				envs = c.Data
				err = nil
				break
			}
		}

		if err != nil && (cmRef.Optional == nil || !*cmRef.Optional) {
			return nil, err
		}
	}

	if envFrom.SecretRef != nil {
		secRef := envFrom.SecretRef
		secret, err := k8sSecretFromSecretManager(secRef.Name, opts.SecretsManager)
		if err == nil {
			for k, v := range secret {
				envs[k] = string(v)
			}
		} else if secRef.Optional == nil || !*secRef.Optional {
			return nil, err
		}
	}

	return envs, nil
}

// envVarValue returns the environment variable value configured within the container's env setting.
// It gets the value from a configMap or secret if specified, otherwise returns env.Value
func envVarValue(env v1.EnvVar, opts *CtrSpecGenOptions) (*string, error) {
	if env.ValueFrom != nil {
		if env.ValueFrom.ConfigMapKeyRef != nil {
			cmKeyRef := env.ValueFrom.ConfigMapKeyRef
			err := fmt.Errorf("cannot set env %v: configmap %v not found", env.Name, cmKeyRef.Name)

			for _, c := range opts.ConfigMaps {
				if cmKeyRef.Name == c.Name {
					if value, ok := c.Data[cmKeyRef.Key]; ok {
						return &value, nil
					}
					err = fmt.Errorf("cannot set env %v: key %s not found in configmap %v", env.Name, cmKeyRef.Key, cmKeyRef.Name)
					break
				}
			}
			if cmKeyRef.Optional == nil || !*cmKeyRef.Optional {
				return nil, err
			}
			return nil, nil
		}

		if env.ValueFrom.SecretKeyRef != nil {
			secKeyRef := env.ValueFrom.SecretKeyRef
			secret, err := k8sSecretFromSecretManager(secKeyRef.Name, opts.SecretsManager)
			if err == nil {
				if val, ok := secret[secKeyRef.Key]; ok {
					value := string(val)
					return &value, nil
				}
				err = fmt.Errorf("secret %v has not %v key", secKeyRef.Name, secKeyRef.Key)
			}
			if secKeyRef.Optional == nil || !*secKeyRef.Optional {
				return nil, fmt.Errorf("cannot set env %v: %v", env.Name, err)
			}
			return nil, nil
		}

		if env.ValueFrom.FieldRef != nil {
			return envVarValueFieldRef(env, opts)
		}

		if env.ValueFrom.ResourceFieldRef != nil {
			return envVarValueResourceFieldRef(env, opts)
		}
	}

	return &env.Value, nil
}

func envVarValueFieldRef(env v1.EnvVar, opts *CtrSpecGenOptions) (*string, error) {
	fieldRef := env.ValueFrom.FieldRef

	fieldPathLabelPattern := `^metadata.labels\['(.+)'\]$`
	fieldPathLabelRegex := regexp.MustCompile(fieldPathLabelPattern)
	fieldPathAnnotationPattern := `^metadata.annotations\['(.+)'\]$`
	fieldPathAnnotationRegex := regexp.MustCompile(fieldPathAnnotationPattern)

	fieldPath := fieldRef.FieldPath

	if fieldPath == "metadata.name" {
		return &opts.PodName, nil
	}
	if fieldPath == "metadata.uid" {
		return &opts.PodID, nil
	}
	fieldPathMatches := fieldPathLabelRegex.FindStringSubmatch(fieldPath)
	if len(fieldPathMatches) == 2 { // 1 for entire regex and 1 for subexp
		labelValue := opts.Labels[fieldPathMatches[1]] // not existent label is OK
		return &labelValue, nil
	}
	fieldPathMatches = fieldPathAnnotationRegex.FindStringSubmatch(fieldPath)
	if len(fieldPathMatches) == 2 { // 1 for entire regex and 1 for subexp
		annotationValue := opts.Annotations[fieldPathMatches[1]] // not existent annotation is OK
		return &annotationValue, nil
	}

	return nil, fmt.Errorf(
		"can not set env %v. Reason: fieldPath %v is either not valid or not supported",
		env.Name, fieldPath,
	)
}

func envVarValueResourceFieldRef(env v1.EnvVar, opts *CtrSpecGenOptions) (*string, error) {
	divisor := env.ValueFrom.ResourceFieldRef.Divisor
	if divisor.IsZero() { // divisor not set, use default
		divisor.Set(1)
	}

	resources, err := getContainerResources(opts.Container)
	if err != nil {
		return nil, err
	}

	var value *resource.Quantity
	resourceName := env.ValueFrom.ResourceFieldRef.Resource
	var isValidDivisor bool

	switch resourceName {
	case "limits.memory":
		value = resources.Limits.Memory()
		isValidDivisor = isMemoryDivisor(divisor)
	case "limits.cpu":
		value = resources.Limits.Cpu()
		isValidDivisor = isCPUDivisor(divisor)
	case "requests.memory":
		value = resources.Requests.Memory()
		isValidDivisor = isMemoryDivisor(divisor)
	case "requests.cpu":
		value = resources.Requests.Cpu()
		isValidDivisor = isCPUDivisor(divisor)
	default:
		return nil, fmt.Errorf(
			"can not set env %v. Reason: resource %v is either not valid or not supported",
			env.Name, resourceName,
		)
	}

	if !isValidDivisor {
		return nil, fmt.Errorf(
			"can not set env %s. Reason: divisor value %s is not valid",
			env.Name, divisor.String(),
		)
	}

	// k8s rounds up the result to the nearest integer
	intValue := int64(math.Ceil(value.AsApproximateFloat64() / divisor.AsApproximateFloat64()))
	stringValue := strconv.FormatInt(intValue, 10)

	return &stringValue, nil
}

func isMemoryDivisor(divisor resource.Quantity) bool {
	switch divisor.String() {
	case "1", "1k", "1M", "1G", "1T", "1P", "1E", "1Ki", "1Mi", "1Gi", "1Ti", "1Pi", "1Ei":
		return true
	default:
		return false
	}
}

func isCPUDivisor(divisor resource.Quantity) bool {
	switch divisor.String() {
	case "1", "1m":
		return true
	default:
		return false
	}
}

func getContainerResources(container v1.Container) (v1.ResourceRequirements, error) {
	result := v1.ResourceRequirements{
		Limits:   v1.ResourceList{},
		Requests: v1.ResourceList{},
	}

	limits := container.Resources.Limits
	requests := container.Resources.Requests

	if limits == nil || limits.Memory().IsZero() {
		mi, err := system.ReadMemInfo()
		if err != nil {
			return result, err
		}
		result.Limits[v1.ResourceMemory] = *resource.NewQuantity(mi.MemTotal, resource.DecimalSI)
	} else {
		result.Limits[v1.ResourceMemory] = limits[v1.ResourceMemory]
	}

	if limits == nil || limits.Cpu().IsZero() {
		result.Limits[v1.ResourceCPU] = *resource.NewQuantity(int64(runtime.NumCPU()), resource.DecimalSI)
	} else {
		result.Limits[v1.ResourceCPU] = limits[v1.ResourceCPU]
	}

	if requests == nil || requests.Memory().IsZero() {
		result.Requests[v1.ResourceMemory] = result.Limits[v1.ResourceMemory]
	} else {
		result.Requests[v1.ResourceMemory] = requests[v1.ResourceMemory]
	}

	if requests == nil || requests.Cpu().IsZero() {
		result.Requests[v1.ResourceCPU] = result.Limits[v1.ResourceCPU]
	} else {
		result.Requests[v1.ResourceCPU] = requests[v1.ResourceCPU]
	}

	return result, nil
}

// getPodPorts converts a slice of kube container descriptions to an
// array of portmapping
func getPodPorts(containers []v1.Container) []types.PortMapping {
	var infraPorts []types.PortMapping
	for _, container := range containers {
		for _, p := range container.Ports {
			if p.HostPort != 0 && p.ContainerPort == 0 {
				p.ContainerPort = p.HostPort
			}
			if p.Protocol == "" {
				p.Protocol = "tcp"
			}
			portBinding := types.PortMapping{
				HostPort:      uint16(p.HostPort),
				ContainerPort: uint16(p.ContainerPort),
				Protocol:      strings.ToLower(string(p.Protocol)),
				HostIP:        p.HostIP,
			}
			// only hostPort is utilized in podman context, all container ports
			// are accessible inside the shared network namespace
			if p.HostPort != 0 {
				infraPorts = append(infraPorts, portBinding)
			}
		}
	}
	return infraPorts
}