User: Jdruker/PE Upgrade/PE DES SAMPLE

Upgrade strategies

<service> supports the following upgrade strategies [templatizable plural and display]:

• Rolling Update
• Blue/Green
• Canary (used in combination with Blue/Green)

The upgrade or rollback process to follow for <service(s)> depends on how you deployed <it/them> initially. Based on the deployment strategy adopted during initial deployment, refer to the corresponding upgrade or rollback section on this page for related instructions.

For more information refer to Upgrade strategies in the Private Edition guide.

Placeholder for optional service-specific content.

Timing

A regular upgrade schedule is necessary (link to info about the s/w versioning policy) but fixes might warrant an earlier upgrade.

Scheduling considerations

When is the best/least disruptive time for actually performing the upgrade, and how long it’ll take. Are there any service dependencies?

Preparatory steps

1. Review the RN to identify changes.
2. Ensure your previously customized values.yaml is available and update it as required to implement changes.
3. Download the new package.

Any other templatizable, generic preparatory steps?

Any other service-specific prep like:

• Recommended backups
• Any permissions-related changes

Rolling Update

Rolling Update: Upgrade

Execute the following command to upgrade <service>:

helm upgrade --install <service> -f <service>values.yaml -n <service package>

Tip: If the only update you need to make to your existing values.yaml is to update the image version, you can pass the image tag as an argument by using the --set flag in the command:

helm upgrade --install <service> -f <service>-values.yaml <service package>  --set <service>.image.tag=<new service version>

Follow the same instructions to upgrade Designer and DAS. For example, the respective commands are:

• Designer:

  helm upgrade --install designer -f designer-values.yaml designer-100.0.112+1401.tgz  --set designer.image.tag=100.0.112.11

• DAS:

  helm upgrade --install designer-das -f designer-das-values.yaml designer-das-100.0.112+1401.tgz  --set das.image.tag=9.0.111.05.5

Rolling Update: Verify the upgrade

Follow usual Kubernetes best practices to verify that the new service version is deployed. See the information about initial deployment for additional functional validation that <service> has upgraded successfully.

Optional additional service-specific info.

Rolling Update: Rollback

1. Update the image version in the <service>.image.tag option in the <service>-values.yaml file.
2. Execute the following command to rollback <service>:

   helm upgrade --install <service> -f <service>-values.yaml <service package>

   Tip: You can also directly pass the image tag as an argument by using the --set flag in the command:

   helm upgrade --install <service> -f <service>-values.yaml <service package>  --set <service>.image.tag=<old service version>

Follow the same instructions to rollback Designer and DAS. For example, the respective commands are:

• Designer:

  helm upgrade --install designer -f designer-values.yaml designer-100.0.112+1401.tgz --set designer.image.tag=100.0.111.05.5

• DAS:

  helm upgrade --install designer-das -f designer-das-values.yaml designer-das-100.0.112+1401.tgz --set das.image.tag=9.0.111.05.5

Verify the rollback

Verify the rollback in the same way that you verified the upgrade (see Rolling Update: Verify the upgrade).

Additional service-specific content.

Blue/Green

Blue/Green: Upgrade Designer

1. Identify the current production color by checking the Designer ingress rules:

   kubectl describe ingress designer-ingress
   

   Green is the production color in the below example as the production host name points to the green service.
   

   

2. Deploy the Designer service on to the non-production color (in this example, blue is the non-production color and assuming the service name is designer-blue):
   

   helm upgrade --install designer-blue -f designer-values.yaml designer-100.0.112+1401.tgz --set designer.deployment.strategy=blue-green --set designer.image.tag=100.0.111.05.5 --set designer.deployment.color=blue

   
   

The non-production color can be accessed with the non-production host name (for example, designer.blue.example.com). Testing can be done using this URL.

NodePort Service

The designer-green release creates a service called designer-green and the designer-blue release creates a service called designer-blue. If you are using NodePort services, ensure that the value of designer.service.nodePort is not the same for both the releases. In other words, you should assign dedicated node ports for the releases. The default value for designer.service.nodePort is 30180. If this was applied to designer-green, use a different value for designer-blue, for example, 30181. Use the below helm command to achieve this:

helm upgrade --install designer-blue -f designer-values.yaml designer-100.0.112+1401.tgz --set designer.deployment.strategy=blue-green --set designer.image.tag=100.0.111.05.5 --set designer.deployment.color=blue --set designer.service.nodePort=30181

Cutover

Once testing is completed on the non-production color, traffic can be moved to the new version by updating the Ingress rules:

• Update the Designer Ingress with the new deployment color by running the following command (in this case, blue is the new deployment color, that is, the non-production color):
  

  helm upgrade --install designer-ingress -f designer-values.yaml designer-100.0.112+1401.tgz --set designer.deployment.strategy=blue-green-ingress --set designer.deployment.color=blue

  
  

Verify the upgrade

• Verify the ingress rules by running the following command:
  kubectl describe ingress designer-ingress
  

  The production host name must point to the new color service, that is, blue.

Blue/Green: Upgrade DAS

1. Identify the current production color by checking the designer-das service selector labels:

   kubectl describe service designer-das</code>.
   #:Green is the production color in the below example as the selector label is <code>color=green

   .
   

   

2. Deploy the DAS service on to the non-production color (in this example, blue is the non-production color and assuming the service name is designer-das-blue):
   

   helm upgrade --install designer-das-blue -f das-values.yaml designer-das-100.0.106+1401.tgz --set das.deployment.strategy=blue-green --set das.image.tag=9.0.111.05.5 --set das.deployment.color=blue

   
   

The non-production color can be accessed with the non-production service name.

Cutover

Once testing is completed on the non-production color, traffic can be moved to the new version by updating the designer-das service.

• Update the designer-das service with the new deployment color (in this example, blue is the new deployment color, that is, non-production color)

  helm upgrade --install designer-das-service -f designer-das-values.yaml designer-das-100.0.106+1401.tgz --set das.deployment.strategy=blue-green-service --set das.deployment.color=blue

Verify the upgrade

• Verify the service by executing the kubectl describe service designer-das command.

  The type label must have the active color's label, that is, color=blue.

Blue/Green: Rollback Designer

To rollback the upgrade, modify the ingress rules to point back to the old deployment pods (green, in this example) by performing a cutover again.

• Perform a cutover using the following command:
  

  helm upgrade --install designer-ingress -f designer-values.yaml designer-100.0.112+1401.tgz --set designer.deployment.strategy=blue-green-ingress --set designer.deployment.color=green

  
  

Verify the rollback

• Verify the rollback in the same way that you verified the upgrade (see Blue-Green: Verify the upgrade).

The type label must have the active color's label, that is, color=green.

Blue/Green: Rollback DAS

To rollback the upgrade, perform a cutover again to point the service back to the old deployment (green).

• Perform a cutover using the following command:
  

  helm upgrade --install designer-das-service -f designer-das-values.yaml designer-das-100.0.106+1401.tgz --set das.deployment.strategy=blue-green-service --set das.deployment.color=green

  
  

Verify the rollback

• Verify the rollback in the same way that you verified the upgrade (see Blue-Green: Verify the upgrade).

The type label must have the active color's label, color=green.

Canary

Canary is optional and is only used along with Blue-Green. It is recommended in production. Canary pods are generally used to test new versions of images with live traffic.

You will not use Canary pods when you are installing the Designer and DAS services for the first time. You will only use Canary pods for testing the new versions when upgrading the services after initial deployment.

For more information on this, see deployment instructions for Canary.

Post-upgrade procedures

Upgrading the Designer workspace

Workspace resources must be upgraded after cutover. Perform the following steps to upgrade the system resources in the Designer workspace:

1. Login to one of the Designer pods using the kubectl exec -it <pod_name > bash command.
2. Execute the following migration command (this creates new directories/new files introduced in the new version):
   node ./bin/cli.js workspace-upgrade -m -t <contact_center_id>
3. Execute the workspace resource upgrade command (this upgrades system resources, such as system service PHP files, internal audio files and callback resources):
   node ./bin/cli.js workspace-upgrade -t <contact_center_id>
   In the above command, contact_center_id , is the Contact Center ID created in GWS for this tenant (workspace resources are located under the Contact Center ID folder (/workspaces/<ccid>/workspace)).

Elasticsearch maintenance recommendations

To help you better manage your indexes and snapshots, and to prevent too many indexes from creating an overflow of shards, Genesys recommends that you set up a scheduled execution of Elasticsearch Curator with the following two actions:

• Delete indexes older than the given threshold according to the index name and mask.
  • sdr-* (3 months)
  • audit-* (12 months)

• Make a snapshot of each index:
  • sdr-* (yesterday and older)
  • audit-*
  • kibana-int-*

Uninstall

In addition to this instruction, are there any preparatory steps (e.g., processes to stop, data to save) and/or service dependencies?

• Execute the following command to uninstall <service>:

  helm uninstall <release-name>