Merge pull request #99530 from shdeshpa07/node-agent-load-affinity

shdeshpa07 · web-flow · commit 83d6181f25b9 · 2025-09-23T17:12:59.000+05:30
Node agent load affinity
diff --git a/backup_and_restore/application_backup_and_restore/installing/installing-oadp-aws.adoc b/backup_and_restore/application_backup_and_restore/installing/installing-oadp-aws.adoc
@@ -54,8 +54,12 @@ include::modules/oadp-installing-dpa-1-3.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agents.adoc[leveloffset=+2]
 include::modules/oadp-configuring-aws-md5sum.adoc[leveloffset=+1]
 include::modules/oadp-configuring-client-burst-qps.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agent-load-affinity.adoc[leveloffset=+1]
+include::modules/oadp-node-agent-load-affinity-guidelines.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-load-concurrency.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-non-root.adoc[leveloffset=+1]
+include::modules/oadp-configuring-repository-maintenance.adoc[leveloffset=+1]
+include::modules/oadp-configuring-velero-load-affinity.adoc[leveloffset=+1]
 include::modules/oadp-configuring-imagepullpolicy.adoc[leveloffset=+1]
 include::modules/oadp-configuring-dpa-multiple-bsl.adoc[leveloffset=+1]
 include::modules/oadp-enabling-csi-dpa.adoc[leveloffset=+2]
diff --git a/backup_and_restore/application_backup_and_restore/installing/installing-oadp-azure.adoc b/backup_and_restore/application_backup_and_restore/installing/installing-oadp-azure.adoc
@@ -39,10 +39,14 @@ include::modules/oadp-using-ca-certificates-with-velero-command.adoc[leveloffset
 // include::modules/oadp-installing-dpa-1-2-and-earlier.adoc[leveloffset=+1]
 include::modules/oadp-installing-dpa-1-3.adoc[leveloffset=+1]
 include::modules/oadp-configuring-client-burst-qps.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agents.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agent-load-affinity.adoc[leveloffset=+1]
+include::modules/oadp-node-agent-load-affinity-guidelines.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-load-concurrency.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-non-root.adoc[leveloffset=+1]
+include::modules/oadp-configuring-repository-maintenance.adoc[leveloffset=+1]
+include::modules/oadp-configuring-velero-load-affinity.adoc[leveloffset=+1]
 include::modules/oadp-configuring-imagepullpolicy.adoc[leveloffset=+1]
-include::modules/oadp-configuring-node-agents.adoc[leveloffset=+2]
 include::modules/oadp-enabling-csi-dpa.adoc[leveloffset=+2]
 include::modules/oadp-about-disable-node-agent-dpa.adoc[leveloffset=+2]
 
diff --git a/backup_and_restore/application_backup_and_restore/installing/installing-oadp-gcp.adoc b/backup_and_restore/application_backup_and_restore/installing/installing-oadp-gcp.adoc
@@ -42,10 +42,14 @@ include::modules/oadp-using-ca-certificates-with-velero-command.adoc[leveloffset
 include::modules/oadp-gcp-wif-cloud-authentication.adoc[leveloffset=+1]
 include::modules/oadp-installing-dpa-1-3.adoc[leveloffset=+1]
 include::modules/oadp-configuring-client-burst-qps.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agents.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agent-load-affinity.adoc[leveloffset=+1]
+include::modules/oadp-node-agent-load-affinity-guidelines.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-load-concurrency.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-non-root.adoc[leveloffset=+1]
+include::modules/oadp-configuring-repository-maintenance.adoc[leveloffset=+1]
+include::modules/oadp-configuring-velero-load-affinity.adoc[leveloffset=+1]
 include::modules/oadp-configuring-imagepullpolicy.adoc[leveloffset=+1]
-include::modules/oadp-configuring-node-agents.adoc[leveloffset=+2]
 include::modules/oadp-enabling-csi-dpa.adoc[leveloffset=+2]
 include::modules/oadp-about-disable-node-agent-dpa.adoc[leveloffset=+2]
 
diff --git a/backup_and_restore/application_backup_and_restore/installing/installing-oadp-ibm-cloud.adoc b/backup_and_restore/application_backup_and_restore/installing/installing-oadp-ibm-cloud.adoc
@@ -26,8 +26,16 @@ include::modules/oadp-setting-resource-limits-and-requests.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agents.adoc[leveloffset=+1]
 // include the module for client burst and qps config
 include::modules/oadp-configuring-client-burst-qps.adoc[leveloffset=+1]
+// include module for load affinity setting
+include::modules/oadp-configuring-node-agent-load-affinity.adoc[leveloffset=+1]
+// include module for load affinity guidelines
+include::modules/oadp-node-agent-load-affinity-guidelines.adoc[leveloffset=+1]
 // include module for loadConcurrency setting
 include::modules/oadp-configuring-node-agent-load-concurrency.adoc[leveloffset=+1]
+// include module for repo maintenance setting
+include::modules/oadp-configuring-repository-maintenance.adoc[leveloffset=+1]
+// include module for velero load affinity setting
+include::modules/oadp-configuring-velero-load-affinity.adoc[leveloffset=+1]
 // include module for image pull policy setting
 include::modules/oadp-configuring-imagepullpolicy.adoc[leveloffset=+1]
 // include the module for configuring multiple BSL
diff --git a/backup_and_restore/application_backup_and_restore/installing/installing-oadp-kubevirt.adoc b/backup_and_restore/application_backup_and_restore/installing/installing-oadp-kubevirt.adoc
@@ -49,10 +49,14 @@ include::modules/oadp-backup-single-vm.adoc[leveloffset=+1]
 include::modules/oadp-restore-single-vm.adoc[leveloffset=+1]
 include::modules/oadp-restore-single-vm-from-multiple-vm-backup.adoc[leveloffset=+1]
 include::modules/oadp-configuring-client-burst-qps.adoc[leveloffset=+1]
-include::modules/oadp-configuring-node-agent-load-concurrency.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-non-root.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agents.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agent-load-affinity.adoc[leveloffset=+1]
+include::modules/oadp-node-agent-load-affinity-guidelines.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agent-load-concurrency.adoc[leveloffset=+1]
+include::modules/oadp-configuring-repository-maintenance.adoc[leveloffset=+1]
+include::modules/oadp-configuring-velero-load-affinity.adoc[leveloffset=+1]
 include::modules/oadp-configuring-imagepullpolicy.adoc[leveloffset=+1]
-include::modules/oadp-configuring-node-agents.adoc[leveloffset=+2]
 include::modules/oadp-incremental-backup-support.adoc[leveloffset=+1]
 
 [IMPORTANT]
diff --git a/backup_and_restore/application_backup_and_restore/installing/installing-oadp-mcg.adoc b/backup_and_restore/application_backup_and_restore/installing/installing-oadp-mcg.adoc
@@ -43,10 +43,14 @@ include::modules/oadp-using-ca-certificates-with-velero-command.adoc[leveloffset
 // include::modules/oadp-installing-dpa-1-2-and-earlier.adoc[leveloffset=+1]
 include::modules/oadp-installing-dpa-1-3.adoc[leveloffset=+1]
 include::modules/oadp-configuring-client-burst-qps.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agents.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agent-load-affinity.adoc[leveloffset=+1]
+include::modules/oadp-node-agent-load-affinity-guidelines.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-load-concurrency.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-non-root.adoc[leveloffset=+1]
+include::modules/oadp-configuring-repository-maintenance.adoc[leveloffset=+1]
+include::modules/oadp-configuring-velero-load-affinity.adoc[leveloffset=+1]
 include::modules/oadp-configuring-imagepullpolicy.adoc[leveloffset=+1]
-include::modules/oadp-configuring-node-agents.adoc[leveloffset=+2]
 include::modules/oadp-enabling-csi-dpa.adoc[leveloffset=+2]
 include::modules/oadp-about-disable-node-agent-dpa.adoc[leveloffset=+2]
 
diff --git a/backup_and_restore/application_backup_and_restore/installing/installing-oadp-ocs.adoc b/backup_and_restore/application_backup_and_restore/installing/installing-oadp-ocs.adoc
@@ -48,10 +48,14 @@ include::modules/oadp-using-ca-certificates-with-velero-command.adoc[leveloffset
 // include::modules/oadp-installing-dpa-1-2-and-earlier.adoc[leveloffset=+1]
 include::modules/oadp-installing-dpa-1-3.adoc[leveloffset=+1]
 include::modules/oadp-configuring-client-burst-qps.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agents.adoc[leveloffset=+1]
+include::modules/oadp-configuring-node-agent-load-affinity.adoc[leveloffset=+1]
+include::modules/oadp-node-agent-load-affinity-guidelines.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-load-concurrency.adoc[leveloffset=+1]
 include::modules/oadp-configuring-node-agent-non-root.adoc[leveloffset=+1]
+include::modules/oadp-configuring-repository-maintenance.adoc[leveloffset=+1]
+include::modules/oadp-configuring-velero-load-affinity.adoc[leveloffset=+1]
 include::modules/oadp-configuring-imagepullpolicy.adoc[leveloffset=+1]
-include::modules/oadp-configuring-node-agents.adoc[leveloffset=+2]
 include::modules/oadp-creating-object-bucket-claim.adoc[leveloffset=+2]
 include::modules/oadp-enabling-csi-dpa.adoc[leveloffset=+2]
 include::modules/oadp-about-disable-node-agent-dpa.adoc[leveloffset=+2]
diff --git a/modules/oadp-configuring-node-agent-load-affinity.adoc b/modules/oadp-configuring-node-agent-load-affinity.adoc
@@ -0,0 +1,63 @@
+// Module included in the following assemblies:
+//
+// * backup_and_restore/application_backup_and_restore/installing/installing-oadp-aws.adoc
+
+:_mod-docs-content-type: PROCEDURE
+[id="oadp-configuring-node-agent-load-affinity_{context}"]
+= Configuring node agent load affinity
+
+[role="_abstract"]
+You can schedule the node agent pods on specific nodes by using the `spec.podConfig.nodeSelector` object of the `DataProtectionApplication` (DPA) custom resource (CR). 
+
+See the following example in which you can schedule the node agent pods on nodes with the label `label.io/role: cpu-1` and `other-label.io/other-role: cpu-2`.
+
+[source,yaml]
+----
+...
+spec:
+  configuration:
+    nodeAgent:
+      enable: true
+      uploaderType: kopia
+      podConfig:
+        nodeSelector:
+          label.io/role: cpu-1
+          other-label.io/other-role: cpu-2
+        ...
+----
+
+You can add more restrictions on the node agent pods scheduling by using the `nodeagent.loadAffinity` object in the DPA spec.
+
+.Prerequisites
+
+* You must be logged in as a user with `cluster-admin` privileges.
+* You have installed the {oadp-short} Operator.
+* You have configured the DPA CR.
+
+.Procedure
+
+* Configure the DPA spec `nodegent.loadAffinity` object as shown in the following example. 
++
+In the example, you ensure that the node agent pods are scheduled only on nodes with the label `label.io/role: cpu-1` and the label `label.io/hostname` matching with either `node1` or `node2`.
++
+[source,yaml]
+----
+...
+spec:
+  configuration:
+    nodeAgent:
+      enable: true
+      loadAffinity: # <1>
+        - nodeSelector:
+            matchLabels:
+              label.io/role: cpu-1
+            matchExpressions: # <2>
+              - key: label.io/hostname
+                operator: In
+                values:
+                  - node1
+                  - node2
+                  ...
+----
+<1> Configure the `loadAffinity` object by adding the `matchLabels` and `matchExpressions` objects.
+<2> Configure the `matchExpressions` object to add restrictions on the node agent pods scheduling.
diff --git a/modules/oadp-configuring-node-agent-load-concurrency.adoc b/modules/oadp-configuring-node-agent-load-concurrency.adoc
@@ -36,6 +36,7 @@ $ oc label node/<node_name> label.io/instance-type='large'
   configuration:
     nodeAgent:
       enable: true
+      uploaderType: kopia
       loadConcurrency:
         globalConfig: 1 # <1>
         perNodeConfig:
diff --git a/modules/oadp-configuring-repository-maintenance.adoc b/modules/oadp-configuring-repository-maintenance.adoc
@@ -0,0 +1,68 @@
+// Module included in the following assemblies:
+//
+// * backup_and_restore/application_backup_and_restore/installing/installing-oadp-aws.adoc
+
+:_mod-docs-content-type: PROCEDURE
+[id="oadp-configuring-repository-maintenance_{context}"]
+= Configuring repository maintenance
+
+[role="_abstract"]
+{oadp-short} repository maintenance is a background job, you can configure it independently of the node agent pods. This means that you can schedule the repository maintenance pod on a node where the node agent is or is not running.
+
+You can use the repository maintenance job affinity configurations in the `DataProtectionApplication` (DPA) custom resource (CR) only if you use Kopia as the backup repository.
+
+You have the option to configure the load affinity at the global level affecting all repositories. Or you can configure the load affinity for each repository. You can also use a combination of global and per-repository configuration.
+
+
+.Prerequisites
+
+* You must be logged in as a user with `cluster-admin` privileges.
+* You have installed the {oadp-short} Operator.
+* You have configured the DPA CR.
+
+.Procedure
+
+* Configure the `loadAffinity` object in the DPA spec by using either one or both of the following methods:
+** Global configuration: Configure load affinity for all repositories as shown in the following example:
++
+[source,yaml]
+----
+...
+spec:
+  configuration:
+    repositoryMaintenance: # <1>
+      global: # <2>
+        podResources:
+          cpuRequest: "100m"
+          cpuLimit: "200m"
+          memoryRequest: "100Mi"
+          memoryLimit: "200Mi"
+        loadAffinity:
+          - nodeSelector:
+              matchLabels:
+                label.io/gpu: 'no'
+              matchExpressions:
+                - key: label.io/location
+                  operator: In
+                  values:
+                    - US
+                    - EU
+----
+<1> Configure the `repositoryMaintenance` object as shown in the example.
+<2> Use the `global` object to configure load affinity for all repositories.
+
+** Per-repository configuration: Configure load affinity per repository as shown in the following example:
++
+[source,yaml]
+----
+...
+spec:
+  configuration:
+    repositoryMaintenance:
+      myrepositoryname: # <1>
+        loadAffinity:
+          - nodeSelector:
+              matchLabels:
+                label.io/cpu: 'yes'
+----
+<1> Configure the `repositoryMaintenance` object for each repository.
diff --git a/modules/oadp-configuring-velero-load-affinity.adoc b/modules/oadp-configuring-velero-load-affinity.adoc
@@ -0,0 +1,57 @@
+// Module included in the following assemblies:
+//
+// * backup_and_restore/application_backup_and_restore/installing/installing-oadp-aws.adoc
+
+:_mod-docs-content-type: PROCEDURE
+[id="oadp-configuring-velero-load-affinity_{context}"]
+= Configuring Velero load affinity
+
+[role="_abstract"]
+With each {oadp-short} deployment, there is one Velero pod and its main purpose is to schedule Velero workloads. To schedule the Velero pod, you can use the `velero.podConfig.nodeSelector` and the `velero.loadAffinity` objects in the `DataProtectionApplication` (DPA) custom resource (CR) spec.
+
+Use the `podConfig.nodeSelector` object to assign the Velero pod to specific nodes. You can also configure the `velero.loadAffinity` object for pod-level affinity and anti-affinity. 
+
+The OpenShift scheduler applies the rules and performs the scheduling of the Velero pod deployment.
+
+
+.Prerequisites
+
+* You must be logged in as a user with `cluster-admin` privileges.
+* You have installed the {oadp-short} Operator.
+* You have configured the DPA CR.
+
+.Procedure
+
+* Configure the `velero.podConfig.nodeSelector` and the `velero.loadAffinity` objects in the DPA spec as shown in the following examples:
+** `velero.podConfig.nodeSelector` object configuration:
++
+[source,yaml]
+----
+...
+spec:
+  configuration:
+    velero:
+      podConfig:
+        nodeSelector:
+          some-label.io/custom-node-role: backup-core
+----
+
+** `velero.loadAffinity` object configuration:
++
+[source,yaml]
+----
+...
+spec:
+  configuration:
+    velero:
+      loadAffinity:
+        - nodeSelector:
+            matchLabels:
+              label.io/gpu: 'no'
+            matchExpressions:
+              - key: label.io/location
+                operator: In
+                values:
+                  - US
+                  - EU
+----
diff --git a/modules/oadp-node-agent-load-affinity-guidelines.adoc b/modules/oadp-node-agent-load-affinity-guidelines.adoc
@@ -0,0 +1,34 @@
+// Module included in the following assemblies:
+//
+// * backup_and_restore/application_backup_and_restore/installing/installing-oadp-aws.adoc
+
+:_mod-docs-content-type: CONCEPT
+[id="oadp-node-agent-load-affinity-guidelines_{context}"]
+= Node agent load affinity guidelines
+
+[role="_abstract"]
+Use the following guidelines to configure the node agent `loadAffinity` object in the `DataProtectionApplication` (DPA) custom resource (CR).
+
+* Use the `spec.nodeagent.podConfig.nodeSelector` object for simple node matching.
+* Use the `loadAffinity.nodeSelector` object without the `podConfig.nodeSelector` object for more complex scenarios.
+* You can use both `podConfig.nodeSelector` and `loadAffinity.nodeSelector` objects, but the `loadAffinity` object must be equal or more restrictive as compared to the `podConfig` object. In this scenario, the `podConfig.nodeSelector` labels must be a subset of the labels used in the `loadAffinity.nodeSelector` object.
+* You cannot use the `matchExpressions` and `matchLabels` fields if you have configured both `podConfig.nodeSelector` and `loadAffinity.nodeSelector` objects in the DPA.
+* See the following example to configure both `podConfig.nodeSelector` and `loadAffinity.nodeSelector` objects in the DPA.
++
+[source,yaml]
+----
+...
+spec:
+  configuration:
+    nodeAgent:
+      enable: true
+      uploaderType: kopia
+      loadAffinity:
+        - nodeSelector:
+            matchLabels:
+              label.io/location: 'US'
+              label.io/gpu: 'no'
+      podConfig:
+        nodeSelector:
+          label.io/gpu: 'no'
+----
