Ver 90388 save restore point before upgrade

Makefile

@@ -389,7 +389,6 @@ docker-build-operator: manifests generate fmt vet ## Build operator docker image
 		--build-arg GO_VERSION=${GO_VERSION} \
 		-f docker-operator/Dockerfile .
 
-
 .PHONY: docker-build-vlogger
 docker-build-vlogger:  ## Build vertica logger docker image
 	docker pull ${VLOGGER_BASE_IMG}:${VLOGGER_ALPINE_VERSION} # Ensure we have the latest alpine version

api/v1/helpers.go

@@ -868,12 +868,24 @@ func (r *RestorePointPolicy) IsValidRestorePointPolicy() bool {
 	return r != nil && r.Archive != "" && ((r.Index > 0) != (r.ID != ""))
 }
 
+// IsValidForSaveRestorePoint returns true id archive name to be used
+// for creating a restore point is set.
+func (r *RestorePointPolicy) IsValidForSaveRestorePoint() bool {
+	return r != nil && r.Archive != ""
+}
+
 // IsRestoreEnabled will return whether the vdb is configured to initialize by reviving from
 // a restore point in an archive
 func (v *VerticaDB) IsRestoreEnabled() bool {
 	return v.Spec.InitPolicy == CommunalInitPolicyRevive && v.Spec.RestorePoint != nil
 }
 
+// IsSaveRestorepointEnabled returns true if the status condition that
+// control restore point is set to true.
+func (v *VerticaDB) IsSaveRestorepointEnabled() bool {
+	return v.IsStatusConditionTrue(SaveRestorePointsNeeded)
+}
+
 // IsHTTPSTLSConfGenerationEnabled return true if the httpstls.json file should
 // be generated by the installer.
 func (v *VerticaDB) IsHTTPSTLSConfGenerationEnabled() (bool, error) {

api/v1/version.go

@@ -83,6 +83,8 @@ const (
 	SandboxSupportedMinVersion = "v24.3.0"
 	// Starting in v24.3.0, we call vclusterops API to get node details instead of executing vsql within the pod
 	FetchNodeDetailsWithVclusterOpsMinVersion = "v24.3.0"
+	// Starting in v24.4.0, nma server will support create restore point to an existing archive
+	SaveRestorePointNMAOpsMinVersion = "v24.4.0"
 )
 
 // GetVerticaVersionStr returns the vertica version, in string form, that is stored

api/v1/verticadb_types.go

@@ -843,6 +843,21 @@ type VerticaDBStatus struct {
 	// +optional
 	// The sandbox statuses
 	Sandboxes []SandboxStatus `json:"sandboxes,omitempty"`
+
+	// +operator-sdk:csv:customresourcedefinitions:type=status
+	// +optional
+	// The details about the last created restore point
+	RestorePoint *RestorePointInfo `json:"restorePoint"`
+}
+
+type RestorePointInfo struct {
+	// +operator-sdk:csv:customresourcedefinitions:type=status
+	// Name of the archive that this restore point was created in.
+	Archive string `json:"archive"`
+	// +operator-sdk:csv:customresourcedefinitions:type=status
+	StartTimestamp string `json:"startTimestamp"`
+	// +operator-sdk:csv:customresourcedefinitions:type=status
+	EndTimestamp string `json:"endTimestamp"`
 }
 
 type SandboxUpgradeState struct {
@@ -890,6 +905,10 @@ const (
 	// VerticaRestartNeeded is a condition that when set to true will force the
 	// operator to stop/start the vertica pods.
 	VerticaRestartNeeded = "VerticaRestartNeeded"
+	// SaveRestorePointsNeeded is a condition that when set to true will,
+	// the operator will create the archive, using the spec.restorePoint.archive
+	// as the archive name to save to.
+	SaveRestorePointsNeeded = "SaveRestorePointsNeeded"
 )
 
 const (

api/v1/verticadb_webhook.go

@@ -193,6 +193,7 @@ func (v *VerticaDB) validateVerticaDBSpec() field.ErrorList {
 	allErrs = v.hasValidSubclusterTypes(allErrs)
 	allErrs = v.hasValidInitPolicy(allErrs)
 	allErrs = v.hasValidRestorePolicy(allErrs)
+	allErrs = v.hasValidSaveRestorePointConfig(allErrs)
 	allErrs = v.hasValidDBName(allErrs)
 	allErrs = v.hasPrimarySubcluster(allErrs)
 	allErrs = v.validateKsafety(allErrs)
@@ -304,6 +305,17 @@ func (v *VerticaDB) hasValidRestorePolicy(allErrs field.ErrorList) field.ErrorLi
 	return allErrs
 }
 
+func (v *VerticaDB) hasValidSaveRestorePointConfig(allErrs field.ErrorList) field.ErrorList {
+	if v.IsSaveRestorepointEnabled() && !v.Spec.RestorePoint.IsValidForSaveRestorePoint() {
+		err := field.Invalid(field.NewPath("spec").Child("restorePoint"),
+			v.Spec.RestorePoint,
+			"restorePoint is invalid. When save restore point is enabled, "+
+				"archive must be specified.")
+		allErrs = append(allErrs, err)
+	}
+	return allErrs
+}
+
 func (v *VerticaDB) validateCommunalPath(allErrs field.ErrorList) field.ErrorList {
 	if v.Spec.InitPolicy == CommunalInitPolicyScheduleOnly {
 		return allErrs

api/v1beta1/verticadb_conversion.go

@@ -261,6 +261,7 @@ func convertToStatus(src *VerticaDBStatus) v1.VerticaDBStatus {
 		Conditions:      make([]metav1.Condition, 0),
 		UpgradeStatus:   src.UpgradeStatus,
 		Sandboxes:       make([]v1.SandboxStatus, len(src.Sandboxes)),
+		RestorePoint:    (*v1.RestorePointInfo)(src.RestorePoint),
 	}
 	for i := range src.Subclusters {
 		dst.Subclusters[i] = convertToSubclusterStatus(src.Subclusters[i])
@@ -285,6 +286,7 @@ func convertFromStatus(src *v1.VerticaDBStatus) VerticaDBStatus {
 		Conditions:      make([]VerticaDBCondition, len(src.Conditions)),
 		UpgradeStatus:   src.UpgradeStatus,
 		Sandboxes:       make([]SandboxStatus, len(src.Sandboxes)),
+		RestorePoint:    (*RestorePointInfo)(src.RestorePoint),
 	}
 	for i := range src.Subclusters {
 		dst.Subclusters[i] = convertFromSubclusterStatus(src.Subclusters[i])

api/v1beta1/verticadb_types.go

@@ -924,6 +924,21 @@ type VerticaDBStatus struct {
 	// +optional
 	// The sandbox statuses
 	Sandboxes []SandboxStatus `json:"sandboxes,omitempty"`
+
+	// +operator-sdk:csv:customresourcedefinitions:type=status
+	// +optional
+	// The details about the last created restore point
+	RestorePoint *RestorePointInfo `json:"restorePoint"`
+}
+
+type RestorePointInfo struct {
+	// +operator-sdk:csv:customresourcedefinitions:type=status
+	// Name of the archive that this restore point was created in.
+	Archive string `json:"archive"`
+	// +operator-sdk:csv:customresourcedefinitions:type=status
+	StartTimestamp string `json:"startTimestamp"`
+	// +operator-sdk:csv:customresourcedefinitions:type=status
+	EndTimestamp string `json:"endTimestamp"`
 }
 
 type SandboxUpgradeState struct {
@@ -974,6 +989,10 @@ const (
 	// VerticaRestartNeeded is a condition that when set to true will force the
 	// operator to stop/start the vertica pods.
 	VerticaRestartNeeded VerticaDBConditionType = "VerticaRestartNeeded"
+	// SaveRestorePointsNeeded is a condition that when set to true will,
+	// the operator will create the archive, using the spec.restorePoint.archive
+	// as the archive name to save to.
+	SaveRestorePointsNeeded VerticaDBConditionType = "SaveRestorePointsNeeded"
 )
 
 // VerticaDBCondition defines condition for VerticaDB

go.mod

@@ -16,7 +16,7 @@ require (
 	github.com/onsi/gomega v1.24.2
 	github.com/pkg/errors v0.9.1
 	github.com/prometheus/client_golang v1.14.0
-	github.com/vertica/vcluster v1.2.1-0.20240824015949-397ced8a6708
+	github.com/vertica/vcluster v1.2.1-0.20240916125930-0654788be326
 	github.com/vertica/vertica-sql-go v1.1.1
 	go.uber.org/zap v1.25.0
 	golang.org/x/text v0.14.0

go.sum

@@ -330,8 +330,8 @@ github.com/theckman/yacspin v0.13.12 h1:CdZ57+n0U6JMuh2xqjnjRq5Haj6v1ner2djtLQRz
 github.com/theckman/yacspin v0.13.12/go.mod h1:Rd2+oG2LmQi5f3zC3yeZAOl245z8QOvrH4OPOJNZxLg=
 github.com/tonglil/buflogr v1.0.1 h1:WXFZLKxLfqcVSmckwiMCF8jJwjIgmStJmg63YKRF1p0=
 github.com/tonglil/buflogr v1.0.1/go.mod h1:yYWwvSpn/3uAaqjf6mJg/XMiAciaR0QcRJH2gJGDxNE=
-github.com/vertica/vcluster v1.2.1-0.20240824015949-397ced8a6708 h1:wjYVhJxvlymSIlBYfRMJuBngPzm/qavKhH/vLcKxzf0=
-github.com/vertica/vcluster v1.2.1-0.20240824015949-397ced8a6708/go.mod h1:yZ3lnvlwuXWgyCZ9Z6bMYisVjP6Ecffl+Lk5wGX/lPQ=
+github.com/vertica/vcluster v1.2.1-0.20240916125930-0654788be326 h1:ZLUnWRhcMQDMDePS+qKZAmQkx/qn/mt6gcrL3y4iaaI=
+github.com/vertica/vcluster v1.2.1-0.20240916125930-0654788be326/go.mod h1:yZ3lnvlwuXWgyCZ9Z6bMYisVjP6Ecffl+Lk5wGX/lPQ=
 github.com/vertica/vertica-sql-go v1.1.1 h1:sZYijzBbvdAbJcl4cYlKjR+Eh/X1hGKzukWuhh8PjvI=
 github.com/vertica/vertica-sql-go v1.1.1/go.mod h1:fGr44VWdEvL+f+Qt5LkKLOT7GoxaWdoUCnPBU9h6t04=
 github.com/yuin/goldmark v1.1.25/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=

pkg/controllers/vdb/offlineupgrade_reconciler.go

@@ -107,6 +107,8 @@
 
 	// Functions to perform when the image changes.  Order matters.
 	funcs := []func(context.Context) (ctrl.Result, error){
+		// Save restore point before start upgrade
+		o.saveRestorePoint,
 		// Initiate an upgrade by setting condition and event recording
 		o.startUpgrade,
 		o.logEventIfThisUpgradeWasNotChosen,
@@ -151,6 +153,15 @@
 	return ctrl.Result{}, o.Manager.logUpgradeSucceeded(sandbox)
 }
 
+func (o *OfflineUpgradeReconciler) saveRestorePoint(ctx context.Context) (ctrl.Result, error) {
+	actor := MakeSaveRestorePointReconciler(o.Rec, o.Vdb, o.Log, o.PFacts, o.Dispatcher, o.Rec.GetClient())
+	saveRestoreRec := actor.(*SaveRestorePoint)
+
+	hostIP, ok := pfacts.FindFirstUpPodIP(true, "")
+	saveRestoreRec.runSaveRestorePointVclusterAPI(ctx, hostIP, archive, pf.sandbox)
+	return o.Manager.startUpgrade(ctx, o.PFacts.GetSandboxName())
+}
+
 func (o *OfflineUpgradeReconciler) startUpgrade(ctx context.Context) (ctrl.Result, error) {
 	return o.Manager.startUpgrade(ctx, o.PFacts.GetSandboxName())
 }