Merge 056426a018 into f4326b50c4

During a replica cluster failover, the designated primary could incorrectly
attempt to restore WALs from its own object store instead of the replica
source, causing recovery to fail. This happened because the previous logic
relied on IsReplica() returning true, but that flag can already be false
while PostgreSQL is still in recovery and needs WALs from the source cluster.

Signed-off-by: Leonardo Cecchi <leonardo.cecchi@enterprisedb.com>

.github/workflows/ci.yml

@@ -44,7 +44,7 @@ jobs:
       - name: Install Dagger
         env:
           # renovate: datasource=github-tags depName=dagger/dagger versioning=semver
-          DAGGER_VERSION: 0.21.7
+          DAGGER_VERSION: 0.21.6
         run: |
           curl -L https://dl.dagger.io/dagger/install.sh | BIN_DIR=$HOME/.local/bin sh
       - name: Run CI task

.github/workflows/release-please.yml

@@ -31,7 +31,7 @@ jobs:
       - name: Install Dagger
         env:
           # renovate: datasource=github-tags depName=dagger/dagger versioning=semver
-          DAGGER_VERSION: 0.21.7
+          DAGGER_VERSION: 0.21.6
         run: |
           curl -L https://dl.dagger.io/dagger/install.sh | BIN_DIR=$HOME/.local/bin sh
       - name: Create image and manifest

.github/workflows/release-publish.yml

@@ -21,7 +21,7 @@ jobs:
       - name: Install Dagger
         env:
           # renovate: datasource=github-tags depName=dagger/dagger versioning=semver
-          DAGGER_VERSION: 0.21.7
+          DAGGER_VERSION: 0.21.6
         run: |
           curl -L https://dl.dagger.io/dagger/install.sh | BIN_DIR=$HOME/.local/bin sh
       - name: Create image and manifest

Taskfile.yml

@@ -206,7 +206,7 @@ tasks:
       - start-build-network
     vars:
       # renovate: datasource=github-tags depName=dagger/dagger versioning=semver
-      DAGGER_VERSION: 0.21.7
+      DAGGER_VERSION: 0.21.6
       DAGGER_ENGINE_IMAGE: registry.dagger.io/engine:v{{ .DAGGER_VERSION }}
     cmds:
       - >

go.mod

@@ -5,7 +5,7 @@ go 1.26.3
 require (
 	github.com/cert-manager/cert-manager v1.20.2
 	github.com/cloudnative-pg/api v1.29.1
-	github.com/cloudnative-pg/barman-cloud v0.5.2-0.20260618100956-53ffbf09b567
+	github.com/cloudnative-pg/barman-cloud v0.5.2-0.20260616073548-263818883a1c
 	github.com/cloudnative-pg/cloudnative-pg v1.29.1
 	github.com/cloudnative-pg/cnpg-i v0.5.0
 	github.com/cloudnative-pg/cnpg-i-machinery v0.4.2
@@ -20,7 +20,7 @@ require (
 	k8s.io/apiextensions-apiserver v0.36.2
 	k8s.io/apimachinery v0.36.2
 	k8s.io/client-go v0.36.2
-	k8s.io/utils v0.0.0-20260617174310-a95e086a2553
+	k8s.io/utils v0.0.0-20260507154919-ff6756f316d2
 	sigs.k8s.io/controller-runtime v0.24.1
 	sigs.k8s.io/kustomize/api v0.21.1
 	sigs.k8s.io/kustomize/kyaml v0.21.1

go.sum

@@ -20,8 +20,8 @@ github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UF
 github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
 github.com/cloudnative-pg/api v1.29.1 h1:FWr8S7EQeOfdhYXyr2cof8wUXLARZiiQt5Qa6ltED7w=
 github.com/cloudnative-pg/api v1.29.1/go.mod h1:QtWF3yzSvIfORMHaSkPAk/o3bhCJwEHJgN3riyRiz3o=
-github.com/cloudnative-pg/barman-cloud v0.5.2-0.20260618100956-53ffbf09b567 h1:vqyRgqs1qDhOO2ZUEx3X4PKJGRnfwXN5xU5KSdjrnps=
-github.com/cloudnative-pg/barman-cloud v0.5.2-0.20260618100956-53ffbf09b567/go.mod h1:vmW7blcg0BECd/9YKqn7NdiF7jz9BlXCWJcGJpzzBCg=
+github.com/cloudnative-pg/barman-cloud v0.5.2-0.20260616073548-263818883a1c h1:5ILvYitKJ3+yT+bDGQM+21QOpa/a8ayCdTqtbWpYJtk=
+github.com/cloudnative-pg/barman-cloud v0.5.2-0.20260616073548-263818883a1c/go.mod h1:vmW7blcg0BECd/9YKqn7NdiF7jz9BlXCWJcGJpzzBCg=
 github.com/cloudnative-pg/cloudnative-pg v1.29.1 h1:ZNEt1TMlnQKXI1kho2UqQuqdfvIvjGln4kN7C1lsmGA=
 github.com/cloudnative-pg/cloudnative-pg v1.29.1/go.mod h1:Sbgx9jVmkle4/gR2U5JHrzDd74sRPOBHDtPkvncg5v8=
 github.com/cloudnative-pg/cnpg-i v0.5.0 h1:/TOzpNT6cwNgrpftTtrnLKdoHgMwd+88vZgXjlVgXeE=
@@ -326,8 +326,8 @@ k8s.io/kube-openapi v0.0.0-20260502001324-b7f5293f4787 h1:kHv8PETbPIVHfqKBYwTNNS
 k8s.io/kube-openapi v0.0.0-20260502001324-b7f5293f4787/go.mod h1:Cyq7UE0QtGe+Zo+/6XFrxiS4Mq0tLyQEONkFzSkfp9o=
 k8s.io/streaming v0.36.2 h1:NSKthPPg9UFSKsRauVJUVGH2Dvn8fhKmY4qrMkw/p98=
 k8s.io/streaming v0.36.2/go.mod h1:z6fV3D+NVkoeqRMtWwlUZK6U17SY/LqNzOxWL6GyR/s=
-k8s.io/utils v0.0.0-20260617174310-a95e086a2553 h1:hmGqDecjc8d7HVzWzRFl0QD9bYuYKbBEG7t8xwnVxfI=
-k8s.io/utils v0.0.0-20260617174310-a95e086a2553/go.mod h1:xDxuJ0whA3d0I4mf/C4ppKHxXynQ+fxnkmQH0vTHnuk=
+k8s.io/utils v0.0.0-20260507154919-ff6756f316d2 h1:wU4tMEhLGgIbLvXQb1cfN+EcM0wf7zC6CPF+C79jroc=
+k8s.io/utils v0.0.0-20260507154919-ff6756f316d2/go.mod h1:xDxuJ0whA3d0I4mf/C4ppKHxXynQ+fxnkmQH0vTHnuk=
 sigs.k8s.io/apiserver-network-proxy/konnectivity-client v0.34.0 h1:hSfpvjjTQXQY2Fol2CS0QHMNs/WI1MOSGzCm1KhM5ec=
 sigs.k8s.io/apiserver-network-proxy/konnectivity-client v0.34.0/go.mod h1:Ve9uj1L+deCXFrPOk1LpFXqTg7LCFzFso6PA48q/XZw=
 sigs.k8s.io/controller-runtime v0.24.1 h1:miPEwrmirImAvgME1L9qebGHrOnGJoVmVdtOU9fRfo4=

internal/cnpgi/common/wal.go

@@ -224,6 +224,7 @@ func (w WALServiceImplementation) Archive(
 }
 
 // Restore implements the WALService interface
+// nolint: gocognit
 func (w WALServiceImplementation) Restore(
 	ctx context.Context,
 	request *wal.WALRestoreRequest,
@@ -238,7 +239,29 @@ func (w WALServiceImplementation) Restore(
 		return nil, err
 	}
 
-	serverName, objectStoreKey := resolveRestoreObjectStore(configuration, w.InstanceName)
+	var serverName string
+	var objectStoreKey types.NamespacedName
+
+	switch {
+	case configuration.Cluster.Status.CurrentPrimary == w.InstanceName &&
+		len(configuration.ReplicaSourceBarmanObjectName) > 0:
+		// restore_command is only called while PostgreSQL is in recovery, so if this
+		// instance is the current primary with a replica source configured, it must be
+		// a designated primary that hasn't completed promotion yet. This covers both
+		// switchover (promotion token) and failover (no promotion token) cases.
+		serverName = configuration.ReplicaSourceServerName
+		objectStoreKey = configuration.GetReplicaSourceBarmanObjectKey()
+
+	case configuration.Cluster.Status.CurrentPrimary == "":
+		// Recovery from object store, using recovery object store
+		serverName = configuration.RecoveryServerName
+		objectStoreKey = configuration.GetRecoveryBarmanObjectKey()
+
+	default:
+		// Using cluster object store
+		serverName = configuration.ServerName
+		objectStoreKey = configuration.GetBarmanObjectKey()
+	}
 
 	var objectStore barmancloudv1.ObjectStore
 	if err := w.Client.Get(ctx, objectStoreKey, &objectStore); err != nil {
@@ -254,33 +277,6 @@ func (w WALServiceImplementation) Restore(
 		ctx, configuration.Cluster, &objectStore, serverName, walName, destinationPath)
 }
 
-// resolveRestoreObjectStore selects the object store and server name to use when
-// restoring a WAL file, based on the role this instance plays in the cluster.
-func resolveRestoreObjectStore(
-	configuration *config.PluginConfiguration,
-	instanceName string,
-) (serverName string, objectStoreKey types.NamespacedName) {
-	switch {
-	case configuration.Cluster.Status.CurrentPrimary == instanceName &&
-		len(configuration.ReplicaSourceBarmanObjectName) > 0:
-		// PostgreSQL never runs restore_command on a live primary; it runs only while
-		// the instance is in recovery (a genuine standby, restoring from a backup, or
-		// being rewound by pg_rewind). So a current primary that still has a replica
-		// source configured can only be a designated primary that has not finished
-		// promoting, and it must keep fetching WAL from the replica source.
-		// Token-agnostic: covers both switchover and failover.
-		return configuration.ReplicaSourceServerName, configuration.GetReplicaSourceBarmanObjectKey()
-
-	case configuration.Cluster.Status.CurrentPrimary == "":
-		// Recovery from object store, using recovery object store
-		return configuration.RecoveryServerName, configuration.GetRecoveryBarmanObjectKey()
-
-	default:
-		// Using cluster object store
-		return configuration.ServerName, configuration.GetBarmanObjectKey()
-	}
-}
-
 func (w WALServiceImplementation) restoreFromBarmanObjectStore(
 	ctx context.Context,
 	cluster *cnpgv1.Cluster,

internal/cnpgi/common/wal_test.go

@@ -1,98 +0,0 @@
-/*
-Copyright © contributors to CloudNativePG, established as
-CloudNativePG a Series of LF Projects, LLC.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-
-SPDX-License-Identifier: Apache-2.0
-*/
-
-package common
-
-import (
-	cnpgv1 "github.com/cloudnative-pg/cloudnative-pg/api/v1"
-	. "github.com/onsi/ginkgo/v2"
-	. "github.com/onsi/gomega"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-
-	"github.com/cloudnative-pg/plugin-barman-cloud/internal/cnpgi/operator/config"
-)
-
-var _ = Describe("resolveRestoreObjectStore", func() {
-	const (
-		namespace = "test-ns"
-		instance  = "cluster-1"
-	)
-
-	// newConfig builds a PluginConfiguration with distinct, recognizable names
-	// for every candidate object store, so each test can assert exactly which
-	// one the routing selected.
-	newConfig := func(currentPrimary, replicaSourceObject string) *config.PluginConfiguration {
-		return &config.PluginConfiguration{
-			Cluster: &cnpgv1.Cluster{
-				ObjectMeta: metav1.ObjectMeta{Namespace: namespace},
-				Status:     cnpgv1.ClusterStatus{CurrentPrimary: currentPrimary},
-			},
-			BarmanObjectName:              "cluster-store",
-			ServerName:                    "cluster-server",
-			RecoveryBarmanObjectName:      "recovery-store",
-			RecoveryServerName:            "recovery-server",
-			ReplicaSourceBarmanObjectName: replicaSourceObject,
-			ReplicaSourceServerName:       "replica-server",
-		}
-	}
-
-	DescribeTable(
-		"selects the correct object store for restoring WAL files",
-		func(cfg *config.PluginConfiguration, wantServer, wantObject string) {
-			gotServer, gotKey := resolveRestoreObjectStore(cfg, instance)
-
-			Expect(gotServer).To(Equal(wantServer))
-			Expect(gotKey.Name).To(Equal(wantObject))
-			Expect(gotKey.Namespace).To(Equal(namespace))
-		},
-
-		// The regression this guards: during a designated-primary promotion the
-		// instance is already the current primary while still in recovery, and it
-		// must pull remaining WALs from the replica source. The routing decision does
-		// not depend on the promotion token, so this single case covers both
-		// switchover and failover.
-		Entry("designated primary in promotion -> replica source",
-			newConfig(instance, "replica-store"),
-			"replica-server", "replica-store"),
-
-		// Guards the len(ReplicaSourceBarmanObjectName) > 0 gate: a current primary
-		// without a barman-backed replica source (plain HA primary, or a replica
-		// cluster whose source is streaming-only) must use the cluster store, not
-		// an empty-named replica source key.
-		Entry("current primary without a replica source -> cluster store",
-			newConfig(instance, ""),
-			"cluster-server", "cluster-store"),
-
-		// Bootstrap / PITR: no current primary yet. Recovery wins even if a replica
-		// source happens to be configured.
-		Entry("no current primary -> recovery store",
-			newConfig("", "replica-store"),
-			"recovery-server", "recovery-store"),
-
-		Entry("ordinary standby -> cluster store",
-			newConfig("cluster-2", ""),
-			"cluster-server", "cluster-store"),
-
-		// A non-primary instance must never route to the replica source, even when
-		// one is configured: only the designated primary catches up from the source.
-		Entry("standby in a replica cluster -> cluster store",
-			newConfig("cluster-2", "replica-store"),
-			"cluster-server", "cluster-store"),
-	)
-})