upgrade: remove unused ValidationTimeout knob

After refactoring, this is no longer used.
The only supported failure strategy is infinite retry, so there's nothing to timeout.

config/operator/crd/bases/scylla.scylladb.com_scyllaclusters.yaml

@@ -1407,10 +1407,7 @@ spec:
                   description: FailureStrategy specifies which logic is executed when upgrade failure happens. Currently only Retry is supported.
                   type: string
                 pollInterval:
-                  description: PollInterval specifies how often upgrade logic polls on state updates. Increasing this value should lower number of requests sent to apiserver, but it may affect overall time required spent during upgrade.
-                  type: string
-                validationTimeout:
-                  description: ValidationTimeout specifies how long it can take for Scylla to boot and enter ready state after upgrade until FailureStrategy is executed.
+                  description: PollInterval specifies how often upgrade logic polls on state updates. Increasing this value should lower number of requests sent to apiserver, but it may affect overall time spent during upgrade.
                   type: string
               type: object
             network:
@@ -1438,7 +1435,7 @@ spec:
                     description: Host to repair, by default all hosts are repaired
                     type: string
                   intensity:
-                    description: Intensity how many token ranges (per shard) to repair in a single Scylla repair job. By default this is 1. If you set it to 0 the number of token ranges is adjusted to the maximum supported by node (see max_repair_ranges_in_parallel in Scylla logs). Valid values are 0 and integers >= 1. Higher values will result in increased cluster load and slightly faster repairs. Changing the intensity impacts repair granularity if you need to resume it, the higher the value the more work on resume. For Scylla clusters that DO NOT SUPPORT ROW-LEVEL REPAIR, intensity can be a decimal between (0,1). In that case it specifies percent of shards that can be repaired in parallel on a repair master node. For Scylla clusters that are row-level repair enabled, setting intensity below 1 has the same effect as setting intensity 1.
+                    description: Intensity how many token ranges (per shard) to repair in a single Scylla repair job. By default this is 1. If you set it to 0 the number of token ranges is adjusted to the maximum supported by node (see max_repair_ranges_in_parallel in Scylla logs). Valid values are 0 and integers >= 1. Higher values will result in increased cluster load and slightly faster repairs. Changing the intensity impacts repair granularity if you need to resume it, the higher the value the more work on resume. For Scylla clusters that *do not support row-level repair*, intensity can be a decimal between (0,1). In that case it specifies percent of shards that can be repaired in parallel on a repair master node. For Scylla clusters that are row-level repair enabled, setting intensity below 1 has the same effect as setting intensity 1.
                     type: string
                   interval:
                     description: Interval task schedule interval e.g. 3d2h10m, valid units are d, h, m, s (default "0").
@@ -1616,7 +1613,7 @@ spec:
                   id:
                     type: string
                   intensity:
-                    description: Intensity how many token ranges (per shard) to repair in a single Scylla repair job. By default this is 1. If you set it to 0 the number of token ranges is adjusted to the maximum supported by node (see max_repair_ranges_in_parallel in Scylla logs). Valid values are 0 and integers >= 1. Higher values will result in increased cluster load and slightly faster repairs. Changing the intensity impacts repair granularity if you need to resume it, the higher the value the more work on resume. For Scylla clusters that DO NOT SUPPORT ROW-LEVEL REPAIR, intensity can be a decimal between (0,1). In that case it specifies percent of shards that can be repaired in parallel on a repair master node. For Scylla clusters that are row-level repair enabled, setting intensity below 1 has the same effect as setting intensity 1.
+                    description: Intensity how many token ranges (per shard) to repair in a single Scylla repair job. By default this is 1. If you set it to 0 the number of token ranges is adjusted to the maximum supported by node (see max_repair_ranges_in_parallel in Scylla logs). Valid values are 0 and integers >= 1. Higher values will result in increased cluster load and slightly faster repairs. Changing the intensity impacts repair granularity if you need to resume it, the higher the value the more work on resume. For Scylla clusters that *do not support row-level repair*, intensity can be a decimal between (0,1). In that case it specifies percent of shards that can be repaired in parallel on a repair master node. For Scylla clusters that are row-level repair enabled, setting intensity below 1 has the same effect as setting intensity 1.
                     type: string
                   interval:
                     description: Interval task schedule interval e.g. 3d2h10m, valid units are d, h, m, s (default "0").

examples/common/operator.yaml

@@ -1422,10 +1422,7 @@ spec:
                   description: FailureStrategy specifies which logic is executed when upgrade failure happens. Currently only Retry is supported.
                   type: string
                 pollInterval:
-                  description: PollInterval specifies how often upgrade logic polls on state updates. Increasing this value should lower number of requests sent to apiserver, but it may affect overall time required spent during upgrade.
-                  type: string
-                validationTimeout:
-                  description: ValidationTimeout specifies how long it can take for Scylla to boot and enter ready state after upgrade until FailureStrategy is executed.
+                  description: PollInterval specifies how often upgrade logic polls on state updates. Increasing this value should lower number of requests sent to apiserver, but it may affect overall time spent during upgrade.
                   type: string
               type: object
             network:
@@ -1453,7 +1450,7 @@ spec:
                     description: Host to repair, by default all hosts are repaired
                     type: string
                   intensity:
-                    description: Intensity how many token ranges (per shard) to repair in a single Scylla repair job. By default this is 1. If you set it to 0 the number of token ranges is adjusted to the maximum supported by node (see max_repair_ranges_in_parallel in Scylla logs). Valid values are 0 and integers >= 1. Higher values will result in increased cluster load and slightly faster repairs. Changing the intensity impacts repair granularity if you need to resume it, the higher the value the more work on resume. For Scylla clusters that DO NOT SUPPORT ROW-LEVEL REPAIR, intensity can be a decimal between (0,1). In that case it specifies percent of shards that can be repaired in parallel on a repair master node. For Scylla clusters that are row-level repair enabled, setting intensity below 1 has the same effect as setting intensity 1.
+                    description: Intensity how many token ranges (per shard) to repair in a single Scylla repair job. By default this is 1. If you set it to 0 the number of token ranges is adjusted to the maximum supported by node (see max_repair_ranges_in_parallel in Scylla logs). Valid values are 0 and integers >= 1. Higher values will result in increased cluster load and slightly faster repairs. Changing the intensity impacts repair granularity if you need to resume it, the higher the value the more work on resume. For Scylla clusters that *do not support row-level repair*, intensity can be a decimal between (0,1). In that case it specifies percent of shards that can be repaired in parallel on a repair master node. For Scylla clusters that are row-level repair enabled, setting intensity below 1 has the same effect as setting intensity 1.
                     type: string
                   interval:
                     description: Interval task schedule interval e.g. 3d2h10m, valid units are d, h, m, s (default "0").
@@ -1631,7 +1628,7 @@ spec:
                   id:
                     type: string
                   intensity:
-                    description: Intensity how many token ranges (per shard) to repair in a single Scylla repair job. By default this is 1. If you set it to 0 the number of token ranges is adjusted to the maximum supported by node (see max_repair_ranges_in_parallel in Scylla logs). Valid values are 0 and integers >= 1. Higher values will result in increased cluster load and slightly faster repairs. Changing the intensity impacts repair granularity if you need to resume it, the higher the value the more work on resume. For Scylla clusters that DO NOT SUPPORT ROW-LEVEL REPAIR, intensity can be a decimal between (0,1). In that case it specifies percent of shards that can be repaired in parallel on a repair master node. For Scylla clusters that are row-level repair enabled, setting intensity below 1 has the same effect as setting intensity 1.
+                    description: Intensity how many token ranges (per shard) to repair in a single Scylla repair job. By default this is 1. If you set it to 0 the number of token ranges is adjusted to the maximum supported by node (see max_repair_ranges_in_parallel in Scylla logs). Valid values are 0 and integers >= 1. Higher values will result in increased cluster load and slightly faster repairs. Changing the intensity impacts repair granularity if you need to resume it, the higher the value the more work on resume. For Scylla clusters that *do not support row-level repair*, intensity can be a decimal between (0,1). In that case it specifies percent of shards that can be repaired in parallel on a repair master node. For Scylla clusters that are row-level repair enabled, setting intensity below 1 has the same effect as setting intensity 1.
                     type: string
                   interval:
                     description: Interval task schedule interval e.g. 3d2h10m, valid units are d, h, m, s (default "0").

pkg/api/v1alpha1/cluster_types.go

@@ -76,9 +76,6 @@ const (
 
 // GenericUpgradeSpec hold generic upgrade procedure parameters.
 type GenericUpgradeSpec struct {
-	// ValidationTimeout specifies how long it can take for Scylla to boot and enter ready state
-	// after image upgrade until FailureStrategy is executed.
-	ValidationTimeout *metav1.Duration `json:"validationTimeout,omitempty"`
 	// FailureStrategy specifies which logic is executed when upgrade failure happens.
 	// Currently only Retry is supported.
 	FailureStrategy GenericUpgradeFailureStrategy `json:"failureStrategy,omitempty"`

pkg/api/v1alpha1/cluster_webhook.go

@@ -46,8 +46,7 @@ var _ webhook.Defaulter = &ScyllaCluster{}
 var _ webhook.Validator = &ScyllaCluster{}
 
 const (
-	DefaultGenericUpgradePollInterval      = time.Second
-	DefaultGenericUpgradeValidationTimeout = 30 * time.Minute
+	DefaultGenericUpgradePollInterval = time.Second
 )
 
 func (c *ScyllaCluster) Default() {
@@ -111,10 +110,6 @@ func (c *ScyllaCluster) Default() {
 		c.Spec.GenericUpgrade.FailureStrategy = GenericUpgradeFailureStrategyRetry
 	}
 
-	if c.Spec.GenericUpgrade.ValidationTimeout == nil {
-		c.Spec.GenericUpgrade.ValidationTimeout = &metav1.Duration{Duration: DefaultGenericUpgradeValidationTimeout}
-	}
-
 	if c.Spec.GenericUpgrade.PollInterval == nil {
 		c.Spec.GenericUpgrade.PollInterval = &metav1.Duration{Duration: DefaultGenericUpgradePollInterval}
 	}

pkg/controllers/cluster/actions/upgrade_version.go

@@ -43,9 +43,8 @@ type ClusterVersionUpgrade struct {
 	ScyllaClient   *scyllaclient.Client
 	ClusterSession cqlSession
 
-	ipMapping         map[string]string
-	pollInterval      time.Duration
-	validationTimeout time.Duration
+	ipMapping    map[string]string
+	pollInterval time.Duration
 
 	currentRack *scyllav1alpha1.RackSpec
 	currentNode *corev1.Pod
@@ -143,15 +142,11 @@ func (a *ClusterVersionUpgrade) Execute(ctx context.Context, s *State) error {
 	a.recorder = s.recorder
 
 	a.pollInterval = scyllav1alpha1.DefaultGenericUpgradePollInterval
-	a.validationTimeout = scyllav1alpha1.DefaultGenericUpgradeValidationTimeout
 
 	if a.Cluster.Spec.GenericUpgrade != nil {
 		if a.Cluster.Spec.GenericUpgrade.PollInterval != nil {
 			a.pollInterval = a.Cluster.Spec.GenericUpgrade.PollInterval.Duration
 		}
-		if a.Cluster.Spec.GenericUpgrade.ValidationTimeout != nil {
-			a.validationTimeout = a.Cluster.Spec.GenericUpgrade.ValidationTimeout.Duration
-		}
 	}
 
 	switch a.upgradeProcedure(ctx) {

pkg/controllers/cluster/actions/upgrade_version_integration_test.go

@@ -54,8 +54,7 @@ var _ = Describe("Cluster controller", func() {
 		BeforeEach(func() {
 			scylla = testEnv.SingleRackCluster(ns)
 			scylla.Spec.GenericUpgrade = &scyllav1alpha1.GenericUpgradeSpec{
-				PollInterval:      &metav1.Duration{Duration: 200 * time.Millisecond},
-				ValidationTimeout: &metav1.Duration{Duration: 5 * time.Second},
+				PollInterval: &metav1.Duration{Duration: 200 * time.Millisecond},
 			}
 
 			Expect(testEnv.Create(ctx, scylla)).To(Succeed())