feat: add --tolerate-failures-status-check flag and deploy.tolerateFa…

…ilures config for improved ci/cd usage

cmd/skaffold/app/cmd/flags.go

@@ -332,6 +332,15 @@ var flagRegistry = []Flag{
 		DefinedOn:     []string{"dev", "debug", "deploy", "run", "apply"},
 		IsEnum:        true,
 	},
+	{
+		Name:          "tolerate-failures-status-check",
+		Usage:         "Configures `status-check` to tolerate failures until Skaffold's statusCheckDeadline duration or the deployments progressDeadlineSeconds  Otherwise deployment failures skaffold encounters will immediately fail the deployment.  Defaults to 'false'",
+		Value:         &opts.TolerateFailuresStatusCheck,
+		DefValue:      false,
+		FlagAddMethod: "BoolVar",
+		DefinedOn:     []string{"dev", "debug", "deploy", "run", "apply"},
+		IsEnum:        true,
+	},
 	{
 		Name:          "fast-fail-status-check",
 		Usage:         "Configures `status-check` to fail immediately if any error occurs.  Otherwise `status-check` will attempt to check all resources once and only then report errors and possibly exit.  Defaults to 'true'",

docs-v2/content/en/schemas/v4beta1.json

@@ -1422,6 +1422,12 @@
           "type": "integer",
           "description": "*beta* deadline for deployments to stabilize in seconds.",
           "x-intellij-html-description": "<em>beta</em> deadline for deployments to stabilize in seconds."
+        },
+        "tolerateFailures": {
+          "type": "boolean",
+          "description": "configures the Skaffold \"status-check\" to tolerate failures (flapping deployments, etc.) until the statusCheckDeadlineSeconds duration or k8s object timeouts such as progressDeadlineSeconds, etc.",
+          "x-intellij-html-description": "configures the Skaffold &quot;status-check&quot; to tolerate failures (flapping deployments, etc.) until the statusCheckDeadlineSeconds duration or k8s object timeouts such as progressDeadlineSeconds, etc.",
+          "default": "false"
         }
       },
       "preferredOrder": [
@@ -1432,6 +1438,7 @@
         "cloudrun",
         "statusCheck",
         "statusCheckDeadlineSeconds",
+        "tolerateFailures",
         "kubeContext",
         "logs"
       ],

docs/content/en/docs/references/cli/_index.md

@@ -139,6 +139,7 @@ Options:
       --status-check=: Wait for deployed resources to stabilize
       --sync-remote-cache='always': Controls how Skaffold manages the remote config cache (see `remote-cache-dir`). One of `always` (default), `missing`, or `never`. `always` syncs remote repositories to latest on access. `missing` only clones remote repositories if they do not exist locally. `never` means the user takes responsibility for updating remote repositories.
       --tail=false: Stream logs from deployed objects
+      --tolerate-failures-status-check=false: Configures `status-check` to tolerate failures until Skaffold's statusCheckDeadline duration or the deployments progressDeadlineSeconds  Otherwise deployment failures skaffold encounters will immediately fail the deployment.  Defaults to 'false'
       --wait-for-connection=false: Blocks ending execution of skaffold until the /v2/events gRPC/HTTP endpoint is hit
 
 Usage:
@@ -168,6 +169,7 @@ Env vars:
 * `SKAFFOLD_STATUS_CHECK` (same as `--status-check`)
 * `SKAFFOLD_SYNC_REMOTE_CACHE` (same as `--sync-remote-cache`)
 * `SKAFFOLD_TAIL` (same as `--tail`)
+* `SKAFFOLD_TOLERATE_FAILURES_STATUS_CHECK` (same as `--tolerate-failures-status-check`)
 * `SKAFFOLD_WAIT_FOR_CONNECTION` (same as `--wait-for-connection`)
 
 ### skaffold build
@@ -453,6 +455,7 @@ Options:
       --sync-remote-cache='always': Controls how Skaffold manages the remote config cache (see `remote-cache-dir`). One of `always` (default), `missing`, or `never`. `always` syncs remote repositories to latest on access. `missing` only clones remote repositories if they do not exist locally. `never` means the user takes responsibility for updating remote repositories.
   -t, --tag='': The optional custom tag to use for images which overrides the current Tagger configuration
       --tail=true: Stream logs from deployed objects
+      --tolerate-failures-status-check=false: Configures `status-check` to tolerate failures until Skaffold's statusCheckDeadline duration or the deployments progressDeadlineSeconds  Otherwise deployment failures skaffold encounters will immediately fail the deployment.  Defaults to 'false'
       --toot=false: Emit a terminal beep after the deploy is complete
       --trigger='notify': How is change detection triggered? (polling, notify, or manual)
       --wait-for-connection=false: Blocks ending execution of skaffold until the /v2/events gRPC/HTTP endpoint is hit
@@ -516,6 +519,7 @@ Env vars:
 * `SKAFFOLD_SYNC_REMOTE_CACHE` (same as `--sync-remote-cache`)
 * `SKAFFOLD_TAG` (same as `--tag`)
 * `SKAFFOLD_TAIL` (same as `--tail`)
+* `SKAFFOLD_TOLERATE_FAILURES_STATUS_CHECK` (same as `--tolerate-failures-status-check`)
 * `SKAFFOLD_TOOT` (same as `--toot`)
 * `SKAFFOLD_TRIGGER` (same as `--trigger`)
 * `SKAFFOLD_WAIT_FOR_CONNECTION` (same as `--wait-for-connection`)
@@ -633,6 +637,7 @@ Options:
       --sync-remote-cache='always': Controls how Skaffold manages the remote config cache (see `remote-cache-dir`). One of `always` (default), `missing`, or `never`. `always` syncs remote repositories to latest on access. `missing` only clones remote repositories if they do not exist locally. `never` means the user takes responsibility for updating remote repositories.
   -t, --tag='': The optional custom tag to use for images which overrides the current Tagger configuration
       --tail=false: Stream logs from deployed objects
+      --tolerate-failures-status-check=false: Configures `status-check` to tolerate failures until Skaffold's statusCheckDeadline duration or the deployments progressDeadlineSeconds  Otherwise deployment failures skaffold encounters will immediately fail the deployment.  Defaults to 'false'
       --toot=false: Emit a terminal beep after the deploy is complete
       --wait-for-connection=false: Blocks ending execution of skaffold until the /v2/events gRPC/HTTP endpoint is hit
       --wait-for-deletions=true: Wait for pending deletions to complete before a deployment
@@ -681,6 +686,7 @@ Env vars:
 * `SKAFFOLD_SYNC_REMOTE_CACHE` (same as `--sync-remote-cache`)
 * `SKAFFOLD_TAG` (same as `--tag`)
 * `SKAFFOLD_TAIL` (same as `--tail`)
+* `SKAFFOLD_TOLERATE_FAILURES_STATUS_CHECK` (same as `--tolerate-failures-status-check`)
 * `SKAFFOLD_TOOT` (same as `--toot`)
 * `SKAFFOLD_WAIT_FOR_CONNECTION` (same as `--wait-for-connection`)
 * `SKAFFOLD_WAIT_FOR_DELETIONS` (same as `--wait-for-deletions`)
@@ -740,6 +746,7 @@ Options:
       --sync-remote-cache='always': Controls how Skaffold manages the remote config cache (see `remote-cache-dir`). One of `always` (default), `missing`, or `never`. `always` syncs remote repositories to latest on access. `missing` only clones remote repositories if they do not exist locally. `never` means the user takes responsibility for updating remote repositories.
   -t, --tag='': The optional custom tag to use for images which overrides the current Tagger configuration
       --tail=true: Stream logs from deployed objects
+      --tolerate-failures-status-check=false: Configures `status-check` to tolerate failures until Skaffold's statusCheckDeadline duration or the deployments progressDeadlineSeconds  Otherwise deployment failures skaffold encounters will immediately fail the deployment.  Defaults to 'false'
       --toot=false: Emit a terminal beep after the deploy is complete
       --trigger='notify': How is change detection triggered? (polling, notify, or manual)
       --wait-for-connection=false: Blocks ending execution of skaffold until the /v2/events gRPC/HTTP endpoint is hit
@@ -803,6 +810,7 @@ Env vars:
 * `SKAFFOLD_SYNC_REMOTE_CACHE` (same as `--sync-remote-cache`)
 * `SKAFFOLD_TAG` (same as `--tag`)
 * `SKAFFOLD_TAIL` (same as `--tail`)
+* `SKAFFOLD_TOLERATE_FAILURES_STATUS_CHECK` (same as `--tolerate-failures-status-check`)
 * `SKAFFOLD_TOOT` (same as `--toot`)
 * `SKAFFOLD_TRIGGER` (same as `--trigger`)
 * `SKAFFOLD_WAIT_FOR_CONNECTION` (same as `--wait-for-connection`)
@@ -1095,6 +1103,7 @@ Options:
       --sync-remote-cache='always': Controls how Skaffold manages the remote config cache (see `remote-cache-dir`). One of `always` (default), `missing`, or `never`. `always` syncs remote repositories to latest on access. `missing` only clones remote repositories if they do not exist locally. `never` means the user takes responsibility for updating remote repositories.
   -t, --tag='': The optional custom tag to use for images which overrides the current Tagger configuration
       --tail=false: Stream logs from deployed objects
+      --tolerate-failures-status-check=false: Configures `status-check` to tolerate failures until Skaffold's statusCheckDeadline duration or the deployments progressDeadlineSeconds  Otherwise deployment failures skaffold encounters will immediately fail the deployment.  Defaults to 'false'
       --toot=false: Emit a terminal beep after the deploy is complete
       --wait-for-connection=false: Blocks ending execution of skaffold until the /v2/events gRPC/HTTP endpoint is hit
       --wait-for-deletions=true: Wait for pending deletions to complete before a deployment
@@ -1153,6 +1162,7 @@ Env vars:
 * `SKAFFOLD_SYNC_REMOTE_CACHE` (same as `--sync-remote-cache`)
 * `SKAFFOLD_TAG` (same as `--tag`)
 * `SKAFFOLD_TAIL` (same as `--tail`)
+* `SKAFFOLD_TOLERATE_FAILURES_STATUS_CHECK` (same as `--tolerate-failures-status-check`)
 * `SKAFFOLD_TOOT` (same as `--toot`)
 * `SKAFFOLD_WAIT_FOR_CONNECTION` (same as `--wait-for-connection`)
 * `SKAFFOLD_WAIT_FOR_DELETIONS` (same as `--wait-for-deletions`)

pkg/skaffold/config/options.go

@@ -32,80 +32,81 @@ type WaitForDeletions struct {
 
 // SkaffoldOptions are options that are set by command line arguments not included in the config file itself
 type SkaffoldOptions struct {
-	Apply                      bool
-	AutoBuild                  bool
-	AutoCreateConfig           bool
-	AutoDeploy                 bool
-	AutoSync                   bool
-	AssumeYes                  bool
-	CacheArtifacts             bool
-	ContainerDebugging         bool
-	Cleanup                    bool
-	DetectMinikube             bool
-	DryRun                     bool
-	EnableRPC                  bool
-	Force                      bool
-	ForceLoadImages            bool
-	IterativeStatusCheck       bool
-	FastFailStatusCheck        bool
-	Notification               bool
-	NoPrune                    bool
-	NoPruneChildren            bool
-	ProfileAutoActivation      bool
-	PropagateProfiles          bool
-	RenderOnly                 bool
-	SkipTests                  bool
-	SkipConfigDefaults         bool
-	Tail                       bool
-	WaitForConnection          bool
-	EnablePlatformNodeAffinity bool
-	EnableGKEARMNodeToleration bool
-	DisableMultiPlatformBuild  bool
-	CheckClusterNodePlatforms  bool
-	MakePathsAbsolute          *bool
-	MultiLevelRepo             *bool
-	CloudRunProject            string
-	CloudRunLocation           string
-	ConfigurationFile          string
-	HydrationDir               string
-	InventoryNamespace         string
-	InventoryID                string
-	InventoryName              string
-	GlobalConfig               string
-	EventLogFile               string
-	RenderOutput               string
-	User                       string
-	CustomTag                  string
-	Namespace                  string
-	CacheFile                  string
-	Trigger                    string
-	KubeContext                string
-	KubeConfig                 string
-	LastLogFile                string
-	DigestSource               string
-	Command                    string
-	MinikubeProfile            string
-	RepoCacheDir               string
-	TransformRulesFile         string
-	VerifyDockerNetwork        string
-	CustomLabels               []string
-	TargetImages               []string
-	Profiles                   []string
-	InsecureRegistries         []string
-	ConfigurationFilter        []string
-	HydratedManifests          []string
-	Platforms                  []string
-	BuildConcurrency           int
-	WatchPollInterval          int
-	StatusCheck                BoolOrUndefined
-	PushImages                 BoolOrUndefined
-	RPCPort                    IntOrUndefined
-	RPCHTTPPort                IntOrUndefined
-	Muted                      Muted
-	PortForward                PortForwardOptions
-	DefaultRepo                StringOrUndefined
-	SyncRemoteCache            SyncRemoteCacheOption
-	WaitForDeletions           WaitForDeletions
+	Apply                       bool
+	AutoBuild                   bool
+	AutoCreateConfig            bool
+	AutoDeploy                  bool
+	AutoSync                    bool
+	AssumeYes                   bool
+	CacheArtifacts              bool
+	ContainerDebugging          bool
+	Cleanup                     bool
+	DetectMinikube              bool
+	DryRun                      bool
+	EnableRPC                   bool
+	Force                       bool
+	ForceLoadImages             bool
+	IterativeStatusCheck        bool
+	FastFailStatusCheck         bool
+	TolerateFailuresStatusCheck bool
+	Notification                bool
+	NoPrune                     bool
+	NoPruneChildren             bool
+	ProfileAutoActivation       bool
+	PropagateProfiles           bool
+	RenderOnly                  bool
+	SkipTests                   bool
+	SkipConfigDefaults          bool
+	Tail                        bool
+	WaitForConnection           bool
+	EnablePlatformNodeAffinity  bool
+	EnableGKEARMNodeToleration  bool
+	DisableMultiPlatformBuild   bool
+	CheckClusterNodePlatforms   bool
+	MakePathsAbsolute           *bool
+	MultiLevelRepo              *bool
+	CloudRunProject             string
+	CloudRunLocation            string
+	ConfigurationFile           string
+	HydrationDir                string
+	InventoryNamespace          string
+	InventoryID                 string
+	InventoryName               string
+	GlobalConfig                string
+	EventLogFile                string
+	RenderOutput                string
+	User                        string
+	CustomTag                   string
+	Namespace                   string
+	CacheFile                   string
+	Trigger                     string
+	KubeContext                 string
+	KubeConfig                  string
+	LastLogFile                 string
+	DigestSource                string
+	Command                     string
+	MinikubeProfile             string
+	RepoCacheDir                string
+	TransformRulesFile          string
+	VerifyDockerNetwork         string
+	CustomLabels                []string
+	TargetImages                []string
+	Profiles                    []string
+	InsecureRegistries          []string
+	ConfigurationFilter         []string
+	HydratedManifests           []string
+	Platforms                   []string
+	BuildConcurrency            int
+	WatchPollInterval           int
+	StatusCheck                 BoolOrUndefined
+	PushImages                  BoolOrUndefined
+	RPCPort                     IntOrUndefined
+	RPCHTTPPort                 IntOrUndefined
+	Muted                       Muted
+	PortForward                 PortForwardOptions
+	DefaultRepo                 StringOrUndefined
+	SyncRemoteCache             SyncRemoteCacheOption
+	WaitForDeletions            WaitForDeletions
 }
 
 type RunMode string

pkg/skaffold/deploy/component/kubernetes/monitor_test.go

@@ -38,6 +38,8 @@ func (m mockStatusConfig) GetKubeContext() string { return "" }
 
 func (m mockStatusConfig) StatusCheckDeadlineSeconds() int { return 0 }
 
+func (m mockStatusConfig) StatusCheckTolerateFailures() bool { return false }
+
 func (m mockStatusConfig) FastFailStatusCheck() bool { return true }
 
 func (m mockStatusConfig) Muted() config.Muted { return config.Muted{} }

pkg/skaffold/kubernetes/status/status_check.go

@@ -78,6 +78,7 @@ type Config interface {
 
 	StatusCheckDeadlineSeconds() int
 	FastFailStatusCheck() bool
+	StatusCheckTolerateFailures() bool
 	Muted() config.Muted
 	StatusCheck() *bool
 }
@@ -89,32 +90,34 @@ type Monitor interface {
 }
 
 type monitor struct {
-	cfg             Config
-	labeller        *label.DefaultLabeller
-	deadlineSeconds int
-	muteLogs        bool
-	failFast        bool
-	seenResources   resource.Group
-	singleRun       singleflight.Group
-	namespaces      *[]string
-	kubeContext     string
-	manifests       manifest.ManifestList
+	cfg              Config
+	labeller         *label.DefaultLabeller
+	deadlineSeconds  int
+	muteLogs         bool
+	failFast         bool
+	tolerateFailures bool
+	seenResources    resource.Group
+	singleRun        singleflight.Group
+	namespaces       *[]string
+	kubeContext      string
+	manifests        manifest.ManifestList
 }
 
 // NewStatusMonitor returns a status monitor which runs checks on selected resource rollouts.
 // Currently implemented for deployments and statefulsets.
 func NewStatusMonitor(cfg Config, labeller *label.DefaultLabeller, namespaces *[]string) Monitor {
 	return &monitor{
-		muteLogs:        cfg.Muted().MuteStatusCheck(),
-		cfg:             cfg,
-		labeller:        labeller,
-		deadlineSeconds: cfg.StatusCheckDeadlineSeconds(),
-		seenResources:   make(resource.Group),
-		singleRun:       singleflight.Group{},
-		namespaces:      namespaces,
-		kubeContext:     cfg.GetKubeContext(),
-		manifests:       make(manifest.ManifestList, 0),
-		failFast:        cfg.FastFailStatusCheck(),
+		muteLogs:         cfg.Muted().MuteStatusCheck(),
+		cfg:              cfg,
+		labeller:         labeller,
+		deadlineSeconds:  cfg.StatusCheckDeadlineSeconds(),
+		seenResources:    make(resource.Group),
+		singleRun:        singleflight.Group{},
+		namespaces:       namespaces,
+		kubeContext:      cfg.GetKubeContext(),
+		manifests:        make(manifest.ManifestList, 0),
+		failFast:         cfg.FastFailStatusCheck(),
+		tolerateFailures: cfg.StatusCheckTolerateFailures(),
 	}
 }
 
@@ -353,7 +356,7 @@ func getStatefulSets(ctx context.Context, client kubernetes.Interface, ns string
 	return resources, nil
 }
 
-func pollResourceStatus(ctx context.Context, cfg kubectl.Config, r *resource.Resource) {
+func pollResourceStatus(ctx context.Context, cfg Config, r *resource.Resource) {
 	pollDuration := time.Duration(defaultPollPeriodInMilliseconds) * time.Millisecond
 	ticker := time.NewTicker(pollDuration)
 	defer ticker.Stop()
@@ -387,7 +390,7 @@ func pollResourceStatus(ctx context.Context, cfg kubectl.Config, r *resource.Res
 			// As any changes to build or deploy dependencies are not triggered, exit
 			// immediately rather than waiting for for statusCheckDeadlineSeconds
 			// TODO: https://github.com/GoogleContainerTools/skaffold/pull/4591
-			if r.HasEncounteredUnrecoverableError() {
+			if r.HasEncounteredUnrecoverableError() && !cfg.StatusCheckTolerateFailures() {
 				r.MarkComplete()
 				return
 			}

pkg/skaffold/runner/runcontext/context.go

@@ -155,6 +155,17 @@ func (ps Pipelines) TransformDenyList() []latest.ResourceFilter {
 	return denylist
 }
 
+func (ps Pipelines) StatusCheckTolerateFailures() bool {
+	failureTolerance := false
+	// set the group status check deadline to maximum of any individually specified value
+	for _, p := range ps.pipelines {
+		if p.Deploy.TolerateFailures {
+			failureTolerance = true
+		}
+	}
+	return failureTolerance
+}
+
 func (ps Pipelines) StatusCheckDeadlineSeconds() int {
 	c := 0
 	// set the group status check deadline to maximum of any individually specified value
@@ -203,6 +214,10 @@ func (rc *RunContext) StatusCheckDeadlineSeconds() int {
 	return rc.Pipelines.StatusCheckDeadlineSeconds()
 }
 
+func (rc *RunContext) StatusCheckTolerateFailures() bool {
+	return rc.Opts.TolerateFailuresStatusCheck || rc.Pipelines.StatusCheckTolerateFailures()
+}
+
 func (rc *RunContext) SkipTests() bool {
 	return rc.Opts.SkipTests
 }