NetBackup™ Deployment Guide for Kubernetes Clusters

Steps to convert from multiple load balancers to none

1. After successfully upgrading the Cloud Scale Technology to 10.5 or later, all the load balancers associated with the media server are upgraded to version 10.5 or later.
2. Deactivate the policy and the schedules. Wait for the running jobs to complete.
3. Add the media2 object in the mediaServers without te network load balancers.
4. Edit the environment, copy the media1 section (remove the load balancer section) and rename it to media2. Save the environment.

   Add media2
   mediaServers:
     - minimumReplicas: 2
       name: media1
       networkLoadBalancer:
         ipList:
         - fqdn: nbux-10-244-33-122.vxindia.veritas.com
           ipAddr: 10.244.33.122
         - fqdn: nbux-10-244-33-123.vxindia.veritas.com
           ipAddr: 10.244.33.123
         type: Private
       nodeSelector:
         labelKey: agentpool
         labelValue: nbuxpool
       paused: false
       replicas: 2
       storage:
         data:
           autoVolumeExpansion: false
           capacity: 50Gi
           storageClassName: managed-csi-hdd
         log:
           autoVolumeExpansion: false
           capacity: 30Gi
           storageClassName: managed-csi-hdd
     - minimumReplicas: 2
       name: media2
       nodeSelector:
         labelKey: agentpool
         labelValue: nbuxpool
       paused: false
       replicas: 2
       storage:
         data:
           autoVolumeExpansion: false
           capacity: 50Gi
           storageClassName: managed-csi-hdd
         log:
           autoVolumeExpansion: false
           capacity: 30Gi
           storageClassName: managed-csi-hdd

5. Once the media2 is successfully added, check the status of the media servers, media2 pods and services. Wait for the all the Pods to come up with media server status as success.

   Get mediaservers
   Kubectl get mediaserver -n <namespace>
   media1 11.0.x-xx 79m nbux-10-244-33-120.vxindia.veritas.com Success
   media2 11.0.x-xx 79m nbux-10-244-33-120.vxindia.veritas.com Success

6. Ensure that all the user settings that were present for media1 pods are added manually to media2 pods also. For example, LogLevel, FIPS Mode, DNAT and so on.
7. Pause both the media reconcilers using the commands:

   Pause mediaServers objects
   kubectl patch -n <namespace> environments <environment name> --type='json' -p='[{"op": "add", "path": "/spec/mediaServers/0/paused", "value": true}]'
   kubectl patch -n <namespace> environments <environment name> --type='json' -p='[{"op": "add", "path": "/spec/mediaServers/1/paused", "value": true}]'
   

   Paused mediaservers
   Kubectl get mediaserver -n <namespace>
   media1 11.0.x-xx 79m nbux-10-244-33-120.vxindia.veritas.com Paused
   media2 11.0.x-xx 79m nbux-10-244-33-120.vxindia.veritas.com Paused

8. Navigate to NetBackup web UI and select Storage. On the Disk Storage, select the Storage server. There are multiple entries for media server with and without the load balancers.
9. Delete the entry for media1. You have to remove all the entries of media servers containing FQDN.
10. Remove the media1 from the CR using the command:

    kubectl patch -n <namespace> environments <environment name> --type='json' -p='[{"op": "remove", "path": "/spec/mediaServers/0"}]'

11. Check the bp.conf file.
12. Delete the media1 entries (FQDN entries of media1) from the bp.conf file. Exec into the primary pod and open /usr/openv/netbackup/bp.conf and delete the entries and save the file.
13. Resume the media2 using the following command:

    kubectl patch -n <namespace> environments <environment name> --type='json' -p='[{"op": "add", "path": "/spec/mediaServers/0/paused", "value": false}]'

14. Note the minimum replica value of media2. Change the minimum replica value using the following command:

    kubectl get pvc -n nbu | grep -i data-media1-media | wc -l 2

    Wait for the pod to be in running state and the status of the media server must be displayed as success.

    Note:

    Although minimum replica value of media server = 0 is supported in cloud scale, for PSF workloads (for example, MongoDB) at least one elastic media server must be running (that is, minimum replica of media server >= 1). The name of media server can be obtained from Host mappings

    For elastic media server, this name is same as the media server pods.

15. Login into the primary server Pod and move the database from previous media server to new media server using the following command:

    bpmedia -movedb -allvolumes -oldserver <old mediaserver name> -newserver <new mediaserver name>

    example: bpmedia -movedb -allvolumes -oldserver nbux-10-244-33-122.vxindia.veritas.com -newserver media2-media-0

    Repeat this step for other media servers.

16. Using the command delete the alias or hostname and repeat this for each media server:

    nbemmcmd -deletehost -machinename <old mediaserver name> -machinetype media

    example: nbemmcmd -deletehost -machinename nbux-10-244-33-122.vxindia.veritas.com -machinetype media

17. Modify the minimum replica value to the original value noted in step 13.
18. Add the entry of the previous media server object alias to the new media server in the Host mappings.

    Note:

    When migrating from multiple load balancers to no load balancers in NetBackup version 11.0 or later, delete the stale host mapping entries as follows:

    Navigate to, NetBackup Web UI ==> Security==> Host mappings ==> Select the Actions menu next to the primary host name ==> Manage mappings ==> Delete the old load balancer media name mapping and click on Save.

19. Restore the backup and check the status in the Activity monitor and to the location where the backup is restored.