Deploy Designer (versions v9012214 and above)

From Genesys Documentation
Revision as of 19:37, March 24, 2022 by MPandian (talk | contribs) (Published)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
This topic is part of the manual Designer Deployment Guide for version Current of Designer.

Contents

Learn how to deploy Designer as a service in a Kubernetes cluster (for DesDepMnfst v9012214 and above).

1. About this document

This document guides you through the process of deploying and configuring Designer and Designer Application Server (DAS) as a service in a Kubernetes (K8s) cluster.

Information on the following topics is provided:

  • Overview of Designer and DAS
  • Configuration details
  • Deployment process
  • Enabling optional features
  • Cleanup
  • Known limitations

1.1 Intended audience

This document is intended for use primarily by system engineers and other members of an implementation team who will be involved in configuring and installing Designer and DAS, and system administrators who will maintain Designer and DAS installations.

To successfully deploy and implement applications in Designer and DAS, you must have a basic understanding of and familiarity with:

  • Network design and operation
  • Network configurations in your organization
  • Kubernetes
  • Genesys Framework architecture and functions

1.2 Before you begin

  1. Install Kubernetes. Refer to the Kubernetes documentation site for installation instructions. You can also refer to the Genesys Docker Deployment Guide for information on Kubernetes and High Availability.
  2. Install Helm according to the instructions outlined in the Helm documentation site.

After you complete the above mandatory procedures, return to this document to complete an on-premise deployment of Designer and DAS as a service in a K8s cluster.

2. Product overview

The following sections provide a brief overview of Designer and DAS.

2.1 Designer

The Designer service provides a web UI to build and manage VXML and SCXML based self-service and assisted service applications for a number of media types. It stores data on the local file system and is synchronized across instances by using services like Network File System (NFS). Genesys customers can build applications using a simple drag and drop method, and assign contact points (Route Points and other media endpoints) to applications directly from the Designer UI. Insights into runtime behavior of applications and troubleshooting aid is provided by Designer Analytics, which includes a rich set of dashboards based on session detail records (SDR) from data stored in Elasticsearch.

Designer offers the following features:

  • Applications for working with phone, chat, email, SMS (text messages), Facebook, Twitter, and open media types.
  • Bots, ASR, TTS capabilities for self-service.
  • Assisted service or routing.
  • Callback.
  • Business Controls.
  • Audio, message management.
  • Grammars management.
  • Contact points management - route points, chat end points, email pop-client/mailboxes.
  • Analytics dashboards through embedded Kibana.

Designer is an Express/Node.js application. The UI is designed using Angular powered Bootstrap. Application data (SCXML and VXML) is stored as a file system. Designer Analytics and Audit data is stored in Elasticsearch.

2.2 Designer Application Server (DAS)

Designer Application Server (DAS) hosts and serves the Designer generated application files (SCXML and VXML), audio, and grammars. It also provides:

  • Runtime evaluation of Business Controls (business hours, special days, emergency flags and data tables).
  • Callback interface to GES.

DAS uses built-in NGINX to front requests. It consists of 3 modules: NGINX, PHP, and Node.js.

  • Requests for static workspace content (SCXML, VXML, JS, audio, grammar, etc) are handled by the NGINX module.
  • Requests for PHP content are processed by the FastCGI PHP module.
  • SDR (Analytics) processing requests are handled by the DAS Node.js module.
Important
Files generated by Designer can be served only by DAS. Designer will work only with DAS.

2.3 Deployment architecture

The below architecture diagram illustrates a sample premise deployment of Designer and DAS:

Premise-Designer-DAS-architecture.png

2.4 High Availability (HA), Disaster Recovery (DR), and Scalability

Designer and DAS must be deployed as highly available in order to avoid single points of failure. A minimum of 2 replicas of each service must be deployed to achieve HA.

The Designer and DAS service pods can be automatically scaled up or down based on metrics such as CPU and memory utilization. The Deployment configuration settings section explains how to configure HA and auto-scaling.

Refer to the Genesys Docker Deployment Guide for more information on general HA recommendation for Kubernetes.

3. Prerequisites

Before deploying Designer, ensure the following resources are deployed, configured, and accessible:

3.1 Mandatory prerequisites

  • Kubernetes 1.12+
  • Helm 3.0
  • Docker
    • To store Designer and DAS docker images to the local docker registry.
  • Ingress Controller
    • If Designer and DAS are accessed from outside of a K8s cluster, it is recommended to deploy/configure an ingress controller (for example, NGINX), if not already available. Also, the Blue-Green deployment strategy works based on the ingress rules.
    • The Designer UI requires Session Stickiness. Configure session stickiness in the annotations parameter in the values.yaml file during Designer installation.
  • Persistent Volumes (PVs)
    • Create persistent volumes for workspace storage (5 GB minimum) and logs (5 GB minimum)
    • Set the access mode for these volumes to ReadWriteMany.
    • The Designer manifest package includes a sample YAML file to create Persistent Volumes required for Designer and DAS.
    • Persistent volumes must be shared across multiple K8s nodes. Genesys recommends using NFS to create Persistent Volumes.
  • Shared file System - NFS
    • For production, deploy the NFS server as highly available (HA) to avoid single points of failure. It is also recommended that the NFS storage be deployed as a Disaster Recovery (DR) topology to achieve continuous availability if one region fails.
    • By Default, Designer and DAS containers run as a Genesys user (uid:gid 500:500). For this reason, the shared volume must have permissions that will allow write access to uid:gid 500:500. The optimal method is to change the NFS server host path to the Genesys user: chown -R genesys:genesys.
    • The Designer manifest package includes a sample YAML file to create an NFS server. Use this only for a demo/lab setup purpose.
    • Azure Files Storage - If you opt for Cloud storage, then Azure Files Storage is an option to consider and has the following requirements:
      A Zone-Redundant Storage for RWX volumes replicated data in zone redundant (check this), shared across multiple pods.
      • Provisioned capacity : 1 TiB
      • Baseline IO/s : 1424
      • Burst IO/s : 4000
      • Egress Rate : 121.4 MiBytes/s
      • Ingress Rate : 81.0 MiBytes/s
  • Genesys Web Services (GWS) 9.x
    • Configure GWS to work with a compatible version of Configuration Server.
  • Other Genesys Components
    • ORS ORS 8.1.400.x
    • Nexus 9.x
    • URS 8.1.400.x

3.2 Optional prerequisites

  • Elasticsearch 7.8.0
    • Elasticsearch is used for Designer Analytics and audit trail.
  • Redis 3.2.x
    • Redis is used for resource index caching and multi-user collaboration locks on Designer resources.

4. Deployment configuration settings (Helm values)

This section provides information on the various settings that have to be configured in Designer and DAS. The configuration settings listed below will be used during the deployment of Designer and DAS. That is, these settings will be used during initial deployment / upgrade. These settings can be configured in the values.yaml Helm file.

4.1 Designer deployment settings

The following table provides information on the Designer deployment settings. These settings are configured in the designer-values.yaml file.

Parameter Description Mandatory? Default Value
designer.deployment.replicaCount Number of service instances to be created. Mandatory 2
designer.deployment.maxreplicaCount The maximum number of replicas to be created. It is recommended to configure this setting if auto-scaling is used. Optional 10
designer.deployment.strategy The deployment strategy to follow. This determines which type of resources are deployed. Valid values are: rollingupdate, blue-green, blue-green-volume, blue-green-ingress, grafana.
  • rollingupdate - default Kubernetes update strategy where resources will be updated using the rolling upgrade strategy.
  • blue-green - for deploying and upgrading the Designer service using the blue-green strategy.
  • blue-green-volume - for the blue/green upgrade, this is to create a Persistent Volume Claim (PVC) for the very first time.
  • blue-green-ingress - for the blue/green upgrade, this is to create an ingress for the first time and update the ingress during a service cutover.
  • grafana - for deploying the Grafana dashboard.
Mandatory rollingupdate
designer.deployment.color This is to deploy/upgrade the Designer service in a blue-green upgrade strategy. Valid values are: blue, green. Optional
designer.deployment.type This is to specify the type of deployment. Valid value: Deployment. Optional Deployment
designer.image.registry The registry that the organization uses for storing images. Mandatory
designer.image.repository Docker repository that contains the images for Designer. Mandatory
designer.image.tag Designer image version. Mandatory 9.0.110.07.7
designer.image.PullPolicy Designer image pull policy (imagePullPolicy). Valid values: Always, IfNotPresent, Never.
  • Always - always pull the image.
  • IfNotPresent - pull the image only if it does not already exist on the node.
  • Never - never pull the image.
Mandatory IfNotPresent
designer.image.imagePullSecrets Secret name containing credentials for authenticating access to the Docker repository. Mandatory
designer.volumes.workspacePvc.create If a persistent volume is to be created, this value has to be true. Mandatory true
designer.volumes.workspacePvc.mountPath The path where the workspace volume is to be mounted inside the Designer container. Mandatory /designer/workspace

(Changing this value is not recommended.)

designer.volumes.workspacePvc.claim Persistent volume claim name for the workspace. Mandatory designer-managed-disk
designer.volumes.workspacePvc.claimSize Size of the persistent volume claim for the workspace.

The persistent volume must be equal to or greater than this size.

Mandatory
designer.volumes.workspacePvc.storageClass storageClassName provided in the persistent volume that is created for the Designer workspace (example, nfs). Mandatory
designer.volumes.logsPvc.create If a PVC volume is to be created, this value has to be true, else false. Mandatory true
designer.volumes.logsPvc.mountPath The path where the Designer logs volume is to be mounted inside the Designer container. Mandatory /designer/logs
designer.volumes.logsPvc.claim Persistent volume claim name for logs. Mandatory designer-logs
designer.volumes.logsPvc.claimSize Size of the persistent volume claim for the Designer logs.

The persistent volume must be equal to or greater than this size.

Mandatory
designer.volumes.logsPvc.storageClass storageClassName provided in the persistent volume that is created for the Designer logs  (example, nfs). Mandatory
designer.podVolumes Log and workspace persistent volume claim names and name of the volumes attached to the pod. Mandatory
designer:
  podVolumes:
    - name: designer-pv-volume
      persistentVolumeClaim:
        claimName: designer-managed-disk
    - name: designer-log-volume
      persistentVolumeClaim:
        claimName: designer-logs
designer.volumeMounts Name and mount path of the volumes to be attached to the Designer pods. Mandatory
volumeMounts:
    - name: designer-pv-volume
      mountPath: /designer/workspace
    - name: designer-log-volume
      mountPath: /designer/logs
designer.livenessProbe.path Designer liveness probe API path. Mandatory /health
designer.livenessProbe.containerPort Port running the container. Mandatory 8888
designer.livenessProbe.startupDelay The liveness probe will be started after a given delay as specified here. Mandatory 20
designer.livenessProbe.checkInterval The interval between each liveness probe request. Mandatory 5
designer.livenessProbe.failureCount Number of liveness probe failures after which, to mark the container as unstable or restart. Mandatory 5
designer.readinessProbe.path Designer readiness probe API path. Mandatory /health
designer.readinessProbe.containerPort Port running the container. Mandatory 8888
designer.readinessProbe.startupDelay The readiness probe will be started after a given delay as specified here. Mandatory 20
designer.readinessProbe.checkInterval The interval between each readiness probe request. Mandatory 5
designer.readinessProbe.failureCount Number of readiness probe failures after which, to mark the container as unstable or restart. Mandatory 5
designer.designerSecrets.enabled This enables providing the GWS Client ID and Secret as an input to the Designer pods. Kubernetes Secrets is used to store the GWS client credentials. Mandatory true
designer.designerSecrets.secrets GWS Client ID and GWS Client Secret. Create a new GWS Client if it does not exist. A link to information on creating a new GWS Client is provided in the Platform settings section. Mandatory
designer.service.enabled Set to true if the service must be created. Optional true
designer.service.type Service type. Valid values are: ClusterIP, NodePort, LoadBalancer. Mandatory NodePort
designer.service.port The Designer service port to be exposed in the cluster. Mandatory 8888
designer.service.targetPort The Designer application port running inside the container. Mandatory http
designer.service.nodePort Port to be exposed in case service type is NodePort. Mandatory for designer.service.type=NodePort. 30180
designer.service.termination_grace_period The period after which Kubernetes starts to delete the pods after service termination. Optional 30 seconds.
designer.ingress.enabled Set to true to enable ingress.

Ingress should be enabled for all cases except for a lab/demo setup.

Mandatory true
designer.ingress.annotations Annotations added for ingress. The Designer UI requires Session Stickiness if the replica count is more than 1. Configure Session Stickiness based on the ingress controller type. Configuration specific to ingress such as Session Stickiness can be provided here. Optional
designer.ingress.paths Ingress path Mandatory [/]
designer.ingress.hosts Hostnames to be configured in ingress for the Designer service. Mandatory - <app1>.example.com

- <app1>.blue.example.com
- <app1>.green.example.com

designer.ingress.tls TLS configuration for ingress. Optional []
designer.resources.limits.cpu Maximum amount of CPU that K8s allocates for the container. Mandatory 600m
designer.resources.limits.memory Maximum amount of memory that K8s allocates for the container. Mandatory 1Gi
designer.resources.requests.cpu Guaranteed CPU allocation for the container. Mandatory 500m
designer.resources.requests.memory Guaranteed memory allocation for the container. Mandatory 512Mi
designer.securityContext.runAsUser This setting controls which user ID the containers are run with. This can be configured to run Designer as a non-root user. You can either use the Genesys user or arbitrary UIDs. Both are supported by the Designer base image. 500 is the ID of the Genesys user.

The file system must reside within the Genesys user account in order to run Designer as a Genesys user. Change the NFS server host path to the Genesys user: chown -R genesys:genesys.

Optional
designer.securityContext.runAsGroup Controls which primary group ID the containers are run with. This can be configured to run Designer as a non-root user. You can either use the Genesys userGroup (GID - 500) or arbitrary GIDs. Both are supported by the Designer base image. Optional
designer.nodeSelector To allow pods to be scheduled based on the labels assigned to the nodes. Optional Default value:

nodeSelector: {}
Sample value:

nodeSelector:
    <label_key>: <label_value>
designer.affinity The K8s standard node affinity and anti-affinity configurations can be added here. Refer to the this topic in the Kubernetes documentation site for sample values. Optional {}
designer.tolerations Tolerations work with taints to ensure that pods are not scheduled on to inappropriate nodes. Refer to the Taints and Tolerations topic in the Kubernetes documentation site for sample values. Optional []
designer.podDisruptionBudget.enabled Set to true if a pod disruption budget is to be created. Optional false
designer.podDisruptionBudget.minAvailable The number of pods that should always be available during a disruption. Optional 1
designer.dnsPolicy The DNS policy that should be applied to the Designer pods. Optional
designer.dnsConfig The DNS configuration that should be applied to the Designer pods. Optional
designer.priorityClassName The priority class name that the pods should belong to. Optional
designer.hpa.enabled Enables K8s Horizontal Pod Autoscaler (HPA). It automatically scales the number of pods based on average CPU utilization and average memory utilization. For more information on HPA refer to this topic in the Kubernetes documentation site. Optional false
designer.hpa.targetCPUPercent The K8s HPA controller will scale up or scale down pods based on the target CPU utilization percentage specified here. It scales up or scales down pods between the range - designer.deployment.replicaCount and designer.deployment.maxreplicaCount. Optional 70
designer.hpa.targetMemoryPercent The K8s HPA controller will scale up or scale down pods based on the target memory utilization percentage specified here. It scales up or scales down pods between the range - designer.deployment.replicaCount and designer.deployment.maxreplicaCount. Optional 70
designer.labels Labels that will be added to the Designer pods. Optional {}
designer.annotations Annotations added to the Designer pods. Optional {}
designer.prometheus.enabled Set to true if Prometheus metrics must be enabled. Optional false
designer.prometheus.tagName Label key assigned to the pods/service to filter out. Optional service
designer.prometheus.tagValue Label value assigned to the pods/service to filter out. Optional designer
designer.prometheus.instance Optional ❴❴instance❵❵
designer.prometheus.serviceMonitor.enabled Set to true<Set/code> if a service monitor resource is needed to monitor the pods through the Kubernetes service. Optional false
designer.prometheus.serviceMonitor.path The path in which the metrics are exposed. Optional /metrics
designer.prometheus.serviceMonitor.interval The scrape interval specified for the Prometheus server. That is, the time interval at which the Prometheus server will fetch metrics from the service. Optional 10s
designer.prometheus.serviceMonitor.labels Labels to be specified for the service monitor resource. Optional
designer.prometheus.alerts.enabled Set to true<Set/code> if Prometheus alerts must to be created. Optional false
designer.prometheus.alerts.customalerts Any custom alerts that are created must be specified here. Optional
designer.prometheus.alerts.labels Labels to be specified for the alerts resource. Optional
designer.prometheus.alerts.<alert_names> Scenarios for which alerts need to be created. Optional designer.prometheus.alerts
containerRestartAlert:
        interval: 3m
        threshold: 5
        AlertPriority: CRITICAL
      MemoryUtilization:
        interval: 1m
        threshold: 70
        AlertPriority: CRITICAL
      endpointAvailable:
        interval: 1m
        AlertPriority: CRITICAL
      CPUUtilization:
        interval: 1m
        threshold: 70
        AlertPriority: CRITICAL
      containerReadyAlert:
        interval: 1m
        readycount: 1
        AlertPriority: CRITICAL
      WorkspaceUtilization:
        interval: 3m
        threshold: 80
        workspaceClaim: designer-managed-disk
        AlertPriority: CRITICAL
      AbsentAlert:
        interval: 1m
        AlertPriority: CRITICAL
      Health:
        interval: 3m
        AlertPriority: CRITICAL
      WorkspaceHealth:
        interval: 3m
        AlertPriority: CRITICAL
      ESHealth:
        interval: 3m
        AlertPriority: CRITICAL
      GWSHealth:
        interval: 3m
        AlertPriority: CRITICAL
designer.grafana.enabled Set to true if the Grafana dashboard is to be created. Optional true
designer.grafana.labels Labels that have to be added to the Grafana ConfigMap. Optional
designer.grafana.annotations Annotations that have to be added to the Grafana ConfigMap. Optional
annotations Enables Kubernetes Annotations and adds it to all the resources that have been created.

For more information, refer to the Annotations topic in the Kubernetes documentation site.

Optional {}
labels Any custom labels can be configured here. It is a key and value pair, for example, key:"value". These labels are added to all resources. Optional {}
podLabels Labels that will be added to all application pods. Optional {}
podAnnotations Annotations that will be added to all application pods. Optional {}

4.1.1 Designer ConfigMap settings

The following table provides information on the environment variables and service-level settings stored in the Designer ConfigMap.

Parameter Description Mandatory? Default Value
designer.designerConfig.create This enables providing environment variables as an input to the Designer pods. It uses a ConfigMap to store the environment variables. Mandatory true
designer.designerConfig.envs.DES_PORT Designer port for container (port in flowsettings.json). Mandatory 8888
designer.designerConfig.envs.DES_APPSERVER_HOST DAS hostname (applicationHost in flowsettings.json). Mandatory das
designer.designerConfig.envs.DES_APPSERVER_PORT DAS port (applicationPort in flowsettings.json). Mandatory 80
designer.designerConfig.envs.DES_DEPLOY_URL This is normally not changed. It is the relative path to the workspace on DAS. The default value /workspaces should be used always (deployURL in flowsettings.json). Mandatory /workspaces
designer.designerConfig.envs.DES_USE_HTCC Set to true so Designer works with GWS. If set to false, Designer defaults to a local mode and may be used temporarily if GWS is unavailable (usehtcc in flowsettings.json). Mandatory false
designer.designerConfig.envs.DES_HTCC_SERVER GWS server host  (htccserver in flowsettings.json). For example, gws.genhtcc.com. Mandatory
designer.designerConfig.envs.DES_HTCC_PORT GWS server port (htccport in flowsettings.json). For example, 80. Mandatory
designer.designerConfig.envs.DES_ENABLE_ANALYTICS To enable or disable Designer Analytics (enableAnalytics in flowsettings.json). Optional false
designer.designerConfig.envs.DES_ES_URL Elasticsearch URL (esUrl in flowsettings.json). For example, http://es-service:9200. Optional
designer.designerConfig.envs.DES_ES_SERVER Elasticsearch Server Host Name (esServer in flowsettings.json). For example, es-service). Optional
designer.designerConfig.envs.DES_ES_PORT Elasticsearch port (esPort in flowsettings.json). For example, 9200. Optional
designer.designerConfig.envs.DES_FILE_LOGGING_ENABLED Enable file logging. If not enabled, Designer will create only verbose logs. Mandatory false
designer.designerFlowSettings.create Set to true to include the contents of the flowsettings.yaml file in a separate ConfigMap. Optional false
designer.designerFlowSettings.envs The flowsettings.yaml file should contain these keys, so that the file's contents will be included in the ConfigMap. Refer to the Updating the flowsettings file section under 9. Post deployment procedures for more information on this. Optional

4.2 DAS deployment settings

The following table provides information on the DAS deployment settings. These settings are configured in the das-values.yaml file. DAS Deployment Settings

Parameter Description Mandatory? Default Value
das.deployment.replicaCount Number of pods to be created. Mandatory 2
das.deployment.maxreplicaCount The maximum number of replicas to be created. It is recommended to configure this setting if auto-scaling is used. Optional 10
das.deployment.strategy The deployment strategy to follow. This determines which type of resources are deployed. Valid values are: rollingupdate, blue-green, blue-green-ingress, blue-green-service, canary.
  • rollingupdate - default Kubernetes update strategy where resources will be updated using the rolling update strategy.
  • blue-green - for deploying and upgrading the DAS service using the blue-green strategy.
  • blue-green-ingress - for the blue-green upgrade, this is to create an ingress for the first time.
  • blue-green-service - for the blue-green upgrade, this is to create a service for the first time, and update the service during a service cutover.
  • canary - to deploy canary pods along with the blue-green pods.
Mandatory rollingupdate
das.deployment.color This is to deploy/upgrade the DAS service using the blue-green upgrade strategy. Valid values are: blue, green. Mandatory for blue-green and blue-green-service strategies.
das.deployment.type Type of Kubernetes controller. Valid value is: Deployment
  • Deployment - if the Designer workspace is stored in the local filesystem (same network where Designer is running) and mounted as NFS.
Optional StatefulSet
das.image.repository Docker repository that contains the images for DAS. Mandatory
das.image.tag DAS image version. Mandatory
das.image.pullPolicy DAS image pull policy (imagePullPolicy). Valid values are: Always, IfNotPresent, Never.
  • Always - always pull the image.
  • IfNotPresent - pull the image only if it does not already exist on the node.
  • Never - never pull the image.
Optional IfNotPresent
das.image.imagePullSecrets Secret name containing the credentials for authenticating access to the Docker repository. Mandatory
das.podVolumes Provides the name of the volume and name of the persistent volume claim to be attached to the pods Mandatory
das:
podVolumes:
- name: workspace
 persistentVolumeClaim:
 claimName: designer-managed-disk
 - name: logs
 persistentVolumeClaim:
 claimName: designer-logs
das.volumes.podPvc.create This volume is usually created to mount a local disk to a DAS container for syncing data in case cloud storage is used for storing Designer files. This value has to be true or false depending on whether the local disk is needed or not Optional false
das.volumes.podPvc.mountPath The path where the workspace volume is to be mounted inside the DAS container. Optional
das.volumes.podPvc.claim Persistent volume claim name for the volume. Optional local-workspace
das.volumes.podPvc.claimSize Size of the persistent volume claim for the pod.

The persistent volume must be equal to or greater than this size.

Optional
das.volumes.podPvc.storageClass storageClassName provided in the persistent volume that is created for DAS (example, nfs). Optional
das.volumes.podPvc.accessModes The read/write priveleges and mount priveleges of the volume claim with respect to the nodes. Valid types are: ReadWriteOnce, ReadOnlyMany, ReadWriteMany.
  • ReadWriteOnce - the volume can be mounted as read-write by a single node.
  • ReadOnlyMany - the volume can be mounted as read-only by many nodes.
  • ReadWriteMany - the volume can be mounted as read-write by many nodes.

For more information, refer to the access modes topic in the Kubernetes documentation site.

Optional ReadWriteOnce
das.volumeMounts The name of the volume and the mount path to be used by the pods. Mandatory
volumeMounts:
- mountPath: /das/www/workspaces
name: workspace
- mountPath: /das/log
name: logs
das.dasSecrets.enabled Set to true if Kubernetes secrets must be created to store keys/credentials/tokens. Optional false
das.dasSecrets.secrets Key and value pairs containing the secrets, such as a username and password. Optional
das.livenessProbe.path DAS liveness probe API path. Mandatory /health
das.livenessProbe.containerPort Port running the container. Mandatory 8081
das.livenessProbe.startupDelay The liveness probe will be started after a given delay as specified here. Mandatory 10
das.livenessProbe.checkInterval The interval between each liveness probe request. Mandatory 5
das.livenessProbe.failureCount Number of liveness probe failures after which, to mark the container as unstable or restart. Mandatory 3
das.readinessProbe.path DAS readiness probe API path. Mandatory /health
das.readinessProbe.containerPort Port running the container. Mandatory 8081
das.readinessProbe.startupDelay The readiness probe will be started after a given delay as specified here. Mandatory 10
das.readinessProbe.checkInterval The interval between each readiness probe request. Mandatory 5
das.readinessProbe.failureCount Number of readiness probe failures after which, to mark the container as unstable or restart. Mandatory 3
das.service.enabled Set to true if the service must be created. Optional true
das.service.type Service type. Valid values are: ClusterIP, NodePort, LoadBalancer. Mandatory NodePort
das.service.port The DAS service port to be exposed in the cluster. Mandatory 80
das.service.targetPort The DAS application port running inside the container. Mandatory http
das.service.nodePort Port to be exposed in case service type is NodePort. Mandatory if das.service.type is NodePort. 30280
das.service.termination_grace_period The period after which Kubernetes starts to delete the pods in case of deletion. Optional 30 seconds.
das.ingress.enabled Set to true to enable ingress.

Ingress should be enabled for all cases except for a lab/demo setup.

Optional false
das.ingress.annotations Annotations added for the ingress resources. Optional
das.ingress.paths Ingress path. Optional [/]
das.ingress.hosts Hostnames to be configured in ingress for the DAS service. Mandatory if ingress is enabled.
das.ingress.tls TLS configuration for ingress. Optional []
das.resources.limits.cpu Maximum amount of CPU that K8s allocates for the container. Mandatory

600m

das.resources.limits.memory Maximum amount of memory that K8s allocates for the container. Mandatory

1Gi

das.resources.requests.cpu Guaranteed CPU allocation for the container. Mandatory 400m
das.resources.requests.memory Guaranteed memory allocation for the container. Mandatory 512Mi
das.securityContext.runAsUser This setting controls which user ID the containers are run with and can be configured to run DAS as a non-root user. You can either use the Genesys user or arbitrary UIDs. Both are supported by the DAS base image. 500 is the ID of the Genesys user.

For more information refer to the Security Context topic in the Kubernetes documentation site.

Optional
das.securityContext.runAsGroup This setting controls which primary group ID the containers are run with and can be configured to run DAS as a non-root user. You can either use the Genesys userGroup (GID - 500) or arbitrary GIDs. Both are supported by the DAS base image. Optional
das.nodeSelector To allow pods to be scheduled based on the labels assigned to the nodes. Optional Default value:

nodeSelector: {} Sample value:

nodeSelector:
    <label_key>: <label_value>
das.affinity The K8s standard node affinity and anti-affinity configurations can be added here. Refer to the this topic in the Kubernetes documentation site for sample values. Optional {}
das.tolerations Tolerations work with taints to ensure that pods are not scheduled on to inappropriate nodes. Refer to the Taints and Tolerations topic in the Kubernetes documentation site for sample values. Optional []
das.podDisruptionBudget.enabled Set to true if a pod disruption budget is to be created. Optional false
das.podDisruptionBudget.minAvailable The number of pods that should always be available during a disruption. Optional 1
das.dnsPolicy The DNS policy that should be applied to the DAS pods. Optional
das.dnsConfig The DNS configuration that should be applied to the DAS pods. Optional
das.priorityClassName The priority class name that the pods should belong to. Optional
das.hpa.enabled Set to true if a K8s Horizontal Pod Autoscaler (HPA) is to be created. Optional false
das.hpa.targetCPUPercent The K8s HPA controller will scale up/down pods based on the target CPU utilization percentage specified. It scale up/down pods between the range deployment.replicaCount to deployment.maxReplicas Optional 75
das.hpa.targetMemoryPercent The K8s HPA controller will scale up or scale down pods based on the target CPU utilization percentage specified here. It scales up or scales down pods between the range - deployment.replicaCount and deployment.maxReplicas. Optional 70
das.labels Labels that will be added to the DAS pods. Optional {}
das.annotations Annotations added to the DAS pods. Optional {}
das.prometheus.enabled Set to true if Prometheus metrics must be enabled. Optional false
das.prometheus.tagName Label key assigned to the pods/service to filter out. Optional service
das.prometheus.tagValue Label key assigned to the pods/service to filter out. Optional designer
das.prometheus.pod Optional ❴❴pod❵❵
das.prometheus.instance Optional ❴❴instance❵❵
das.prometheus.serviceMonitor.enabled Set to true<Set/code> if a service monitor resource is needed to monitor the pods through the Kubernetes service. Optional false
das.prometheus.serviceMonitor.path The path in which the metrics are exposed. Optional /metrics
das.prometheus.serviceMonitor.interval The scrape interval specified for the Prometheus server. That is, the time interval at which the Prometheus server will fetch metrics from the service. Optional 10s
das.prometheus.serviceMonitor.labels Labels to be specified for the service monitor resource. Optional
das.prometheus.alerts.enabled Set to true<Set/code> if Prometheus alerts must to be created. Optional false
das.prometheus.alerts.labels Labels to be specified for the alerts resource. Optional
das.prometheus.alerts.customalerts Any custom alerts that are created must be specified here. Optional
das.prometheus.alerts.<alert_names> Scenarios for which alerts need to be created. Optional

das.prometheus.alerts.<alert_name>

containerRestartAlert:
        interval: 3m
        threshold: 5
        AlertPriority: CRITICAL
      MemoryUtilization:
        interval: 1m
        threshold: 75
        AlertPriority: CRITICAL
      endpointAvailable:
        interval: 1m
        AlertPriority: CRITICAL
      CPUUtilization:
        interval: 1m
        threshold: 75
        AlertPriority: CRITICAL
      containerReadyAlert:
        interval: 5m
        readycount: 1
        AlertPriority: CRITICAL
      rsyncContainerReadyAlert:
        interval: 5m
        readycount: 1
        AlertPriority: CRITICAL
      WorkspaceUtilization:
        interval: 3m
        threshold: 70
        workspaceClaim: designer-managed-disk
        AlertPriority: CRITICAL
      AbsentAlert:
        interval: 1m
        AlertPriority: CRITICAL
      LocalWorkspaceUtilization:
        interval: 3m
        threshold: 70
        AlertPriority: CRITICAL
      Health:
        interval: 3m
        AlertPriority: CRITICAL
      WorkspaceHealth:
        interval: 3m
        AlertPriority: CRITICAL
      PHPHealth:
        interval: 3m
        AlertPriority: CRITICAL
      ProxyHealth:
        interval: 3m
        AlertPriority: CRITICAL
      PhpLatency:
        interval: 1m
        threshold: 10
        AlertPriority: CRITICAL
      HTTPLatency:
        interval: 1m
        threshold: 60
        AlertPriority: CRITICAL
      HTTP4XXCount:
        interval: 5m
        threshold: 100
        AlertPriority: CRITICAL
      HTTP5XXCount:
        interval: 5m
        threshold: 100
        AlertPriority: CRITICAL
das.grafana.enabled Set to true if the Grafana dashboard is to be created. Optional true
das.grafana.labels Labels that must be added to the Grafana ConfigMap. Optional
das.grafana.annotations Annotations that must be added to the Grafana ConfigMap. Optional
annotations Enables Kubernetes Annotations and adds it to all the resources that have been created.

For more information, refer to the Annotations topic in the Kubernetes documentation site.

Optional {}
labels Any custom labels can be configured here. It is a key and value pair, for example, key:"value". These labels are added to all resources. Optional {}
podLabels Labels that will be added to all application pods. Optional {}
podAnnotations Annotations that will be added to all application pods. Optional {}

4.2.1 DAS ConfigMap settings

Parameter Description Mandatory? Default Value
das.dasConfig.create This setting enables providing environment variables as an input to the DAS pods. It uses a ConfigMap to store the environment variables. Mandatory true
das.dasConfig.envs.DAS_FILE_LOGGING_ENABLED Enables file logging. DAS supports only std out logging. This should always be set to false. Mandatory false
das.dasConfig.envs.DAS_LOG_LEVEL Enables log levels. Valid values are: FATAL, ERROR, WARN, INFO, DEBUG, TRACE. Optional DEBUG
das.dasConfig.envs.DAS_STDOUT_LOGGING_ENABLE Enables standard output console logging. Mandatory true
das.dasConfig.envs.DAS_SERVICES_ELASTICSEARCH_ENABLED To enable Designer Analytics. This configuration is required for DAS to initialize ES templates. Optional false
das.dasConfig.envs.DAS_SERVICES_ELASTICSEARCH_HOST Elasticsearch server host name with http:// prefix. For example, http://es-service. Optional
das.dasConfig.envs.DAS_SERVICES_ELASTICSEARCH_PORT Elasticsearch port. For example, 80. Optional

5. Post deployment Designer configuration settings

Post deployment, Designer configuration is managed from the following 3 locations:

5.1 Flow settings

Flow Settings is used for controlling global Designer settings that are applicable to all tenants and it contains bootstrap configuration settings such as port, GWS info, and DAS URL.

Configuration path - /workspace/designer/flowsettings.json.

This will be configured using the helm install. Refer to the Updating the flowsettings file section under 9. Post deployment procedures for more information on updating the flowsettings.json file.

5.2 Tenant settings

These are tenant specific settings if the Designer service is configured with multi-tenancy .

Configuration  path - workspace/<contactcenter-Id>/config/tenantsettings.json.

The user should logout and log back in after any changes to the tenantsettings.json file. The Designer UI will continue to show the older features until the user logs out and logs back in.

Tenant specific settings are configured by directly editing the file in the above path.

5.3 DesignerEnv transaction list

The DesignerEnv transaction list is available in Configuration Server (Tenant/Transactions/DesignerEnv). This is mostly used to control the run-time settings. Any change to the DesignerEnv transaction list does not require the application to be published again or a new build for the application.

The user should log out and log back in for the changes to reflect in the Designer UI.

The DesignerEnv transaction list is configured using Agent Setup.

5.4 Post deployment configuration settings reference table

Category: Analytics
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
enableAnalytics
(optional)
Yes Yes No This flag enables or disables the analytics feature.

Sample value: true
Default value: false

esUrl
(optional)
Yes Yes No Elasticsearch URL

Sample value: http://es-spot.usw1.genhtcc.com:80

esServer
(optional)
Yes Yes No Elasticsearch server host name (for example, es-service). Sample value: es-spot.usw1.genhtcc.com
esPort
(optional)
Yes Yes No Elasticsearch port. Sample value: 80
ReportingURL
(optional)
No No Yes

Section: reporting

URL of Elasticsearch where Designer applications will report data.

Sample value: http://es-spot.usw1.genhtcc.com:80

esMaxQueryDuration
(optional)
Yes Yes No The maximum time range (in days) to query in Designer Analytics. Each day's data is stored in a separate index in Elasticsearch.

Sample value: 90
Default value: 90

sdrMaxObjCount
(optional)
Yes Yes No The maximum count of nested type objects that will be captured in SDRs. When set to -1, which is the default value, no objects will be trimmed. All the milestones or activities visited in runtime are expected to be captured in an SDR.

Sample value: 20

SdrTraceLevel
(optional)
Yes Yes No Value are:
  • 100 — Debug level and up. Currently, there are no Debug messages.
  • 200 — Standard level and up. This setting will show all blocks that are entered during a call in the blocks array.
  • 300 — Important level and up. This setting filters out all blocks from the blocks array, except those containing data that will change from call to call (such as the Menu block and User Input block).

Sample value: 300
Default value: 300

Category: Audit
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
enableESAuditLogs
(optional)
Yes Yes No Enable or disable audit logs captured in Elasticsearch.

Sample value: false
Default value: false

enableFSAuditLogs
(optional)
Yes Yes No Enable or Disable audit logs captured in the file system under the logs directory or in standard output.

Sample value: true
Default value: true

maxAppSizeCompare
(optional)
Yes Yes No The maximum size of data object for which a difference will be captured in the audit logs, value in bytes. That is, the difference between the Designer object's old value and new value.

Sample value: 1000000
Default value: 1000000

enableReadAuditLogs
(optional)
Yes Yes No Control whether reading of Designer objects is captured in audit trails. If enabled any Designer object viewed in the UI will be recorded in the audit logs.

Sample value: false
Default value: false

Category: Authorization
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
disableRBAC
(optional)
Yes Yes No Controls if Designer reads and enforces permissions associated with the logged in user's roles.

Sample value: false
Default value: false

rbacSection
(optional)
Yes Yes No In a Role object, the name of the section within the Annex where the privileges are stored.

Sample value: CfgGenesysAdministratorServer
Default value: CfgGenesysAdministratorServer

disablePBAC
(optional)
Yes Yes No Controls if Designer allows partitioning of the Designer workspace and restricts a user's access to Designer objects in the user's partitions.

Sample value: false
Default value: false

Category: Collaboration
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
locking
(optional)
Yes No No The type of locking used, in an editing session for applications, modules, or data tables. Valid values are: file, redis, none.
  • none - resources are not locked and can be edited simultaneously by multiple users which can result in one user overwriting another user's changes.
  • file - uses files to keep track of locks and relies on shared storage (for example, NFS) to make lock files available to each Designer pod. Lock files are stored in the same location as the user's Designer workspace.
  • redis - uses Redis for storing resource locks and is recommended for production environments.

Sample value: file
Default value: file

Category: DAS
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
applicationHost
(mandatory)
Yes No No The server name Designer uses to generate the URL to the application. ORS and MCP fetch the application code and other resources from this URL.

Sample value: das.usw1.genhtcc.com
Default value: localhost

applicationPort Yes No No The corresponding port to be used with applicationHost.

Sample value: 80
Default value: 80

deployURL Yes No No This is normally not changed. It is the relative path to the workspace on DAS.

Sample value: /workspace
Default value: /workspace

Category: Digital
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
rootsSRL
(optional)
Yes Yes No If specified, this is used to filter which Root Categories to display when selecting Standard Responses.

Sample value: Any REGular EXpression (REGEX).

maxFlowEntryCount
(optional)
Yes No Yes

Section: flowsettings

Specify how many times the same application can process a specific digital interaction.

Sample value: 20
Default value: 20

Category: External APIs
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
httpProxy
(optional)
Yes Yes Yes

Secion: flowsettings

Specify the proxy used for external requests and nexus API calls (if enable_proxy is true).

Sample value: [http://vpcproxy-000-int.geo.genprim.com:8080

redundantHttpProxy
(optional)
Yes Yes Yes

Section: flowsettings

Specify the backup proxy used for external requests and nexus API calls (if enable_proxy is true), when httpProxy is down.

Sample value: [http://vpcproxy-001-int.geo.genprim.com:8080

Category: Features
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
features Yes Yes No This is an object. See the 5.5 Features section for a list of supported features.

Default value:

{
  nexus: true,
  enableBulkAudioImport: true
}
Category: GWS
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
usehtcc Yes No No Set to true so that Designer works with GWS. If set to false, Designer defaults to a local mode and may be used temporarily if GWS is unavailable.

Sample value: true
Default value: false

htccServer Yes No No GWS Server

Sample value: gws-usw1-int.genhtcc.com
Default value: gws-usw1-int.genhtcc.com

htccport Yes No No GWS port.

Sample value: 80
Default value: 80

ssoLoginUrl Yes No No URL of GWS authentication UI. Designer redirects to this URL for authentication.

Sample value: https://gws-usw1.genhtcc.com
Default value: https://gws-usw1.genhtcc.com

maxConcurrentHTCCRequest
(optional)
Yes No No For batch operations to GWS, the max number of concurrent requests that Designer will send to GWS.

Sample value: 5
Default value: 5

batchOperationResultTTL
(optional)
Yes No No For batch operations to GWS, the time, in milliseconds, for which duration Designer stores the results of a batch operation on the server, before deleting them.

Sample value: 100000
Default value: 100000

Category: Help
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
docsMicroserviceURL
(optional)
Yes No No URL for Designer documentation.

Default value: https://docs.genesys.com/Documentation/PSAAS/Public/Administrator/Designer

Category: IVR
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
recordingType
(optional)
Yes Yes No Specify the recording type to be used in Record block. Set as GIR. If the option is missing or blank, Full Call Recording type will be used.

Sample value: GIR
Default value: GIR

Category: Logging
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
logging: {
designer: { level:
debug },
audit: { level: trace},
auditdebug: { level: debug },
cli: { level: debug }
}

(optional)

Yes No No Specify Designer log levels. Each field has valid values: trace, debug, info, warn, error, or fatal.
  • designer - log level of Designer.
  • audit - log level of audit.
  • auditdebug - log level of audit debug, this will log detailed audit information.
  • cli - log level for cli commands executed on Designer.

Sample value:

logging: {
designer: { level: debug},
audit: { level: trace },
auditdebug: { level: debug},
cli: { level: debug }
}

Default value:
logging: {
designer: { level: debug },
audit: { level: trace },
auditdebug: { level: debug },
cli: { level: debug }
}
Category: Nexus
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
url
(optional)
No No Yes

Section: nexus

URL of Nexus that typically includes the API version path. For example, https://nexus-server/nexus/api/v3.

Default value: http://nex-dev.usw1.genhtcc.com

password
(optional)
No No Yes

Section: nexus

The Nexus x-api-key created by Nexus deployment.

Default value: dc4qeiro13nsof569dfn234smf

enable_proxy
(optional)
No No Yes

Section: nexus

Boolean value to indicate if httpProxy is used to reach Nexus.

Default value: false

profile
(optional)
No No Yes

Section: nexus

Enable Contact Identification via Nexus (for example, to enable Last Called Agent routing).
Category: Process
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
port Yes No No Designer process port in the container. Normally, the default value should be left as is.

Sample value: 8888
Defualt value: 3000

Category: Provisioning
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
primarySwitch
(optional)
Yes Yes No Specify the primary switch name if more than one switch is defined for the tenant. Designer fetches and works with route points from this switch.

Default value: us-west-1

Category: Routing
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
ewtRefreshTimeout
(optional)
NO No Yes

Section: flowsettings

Specify the interval (in seconds) at which to refresh the Estimated Waiting Time when routing an interaction.

Sample value: 5
Default value: 1

Category: Redis
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
redis: {
host: "",
port: "",
tlsEnabled: true,
lockTimeout: 120,
listTimeout: 1800
}

(optional)
Yes No No Used by Designer for resource index caching and multi-user collaboration locks on Designer resources.

It is a separate object that contains:

  • host - Redis host name.
  • port - Redis port.
  • tlsEnabled - TLS enabled or not.
  • lockTimeout - Timeout, in seconds, before a resource lock is released for an editing session of applications, modules, or data tables.
  • listTimeout - The cache expiry timeout (in seconds) of the application list and shared modules list. By default, it is 30 minutes. That is, any new application/modules created in the UI will be seen in the listing page after 30 mins. It can be reduced to a smaller value. This is to improve the page loading performance of the Applications and Shared Modules page. A better performance is achieved with a higher value.

Sample value:

redis: {
host: "",
port: "",
tlsEnabled: true,
lockTimeout: 120,
listTimeout: 1800
}

Default value:
redis: {
host: redis.server.genhtcc.com,
port: 6379,
tlsEnabled: true,
lockTimeout: 120,
listTimeout: 1800
}
Category: Security
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
zipFileSizeLimitInMegaBytes
(optional)
Yes Yes No Defines the maximum zipFile size limit (in megabytes) during bulk audio import.

Sample value: 50

disableCSRF
(optional)
Yes Yes No Disable CSRF attack protection. For more information, refer to this topic in the CWE site.

By default, CSRF attack protection is enabled. It can be disabled by setting this flag to true.

Sample value: false
Default value: false

disableSecureCookie
(optional)
Yes No No Disables the secure cookies header.

Sample value: false
Default value: false

Category: Session
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
idleTimeout
(optional)
Yes Yes No Idle timeout, in seconds, before a user session is terminated while editing applications, modules, or data tables.

Sample value: 840
Default value: 840

lockTimeout
(optional)
Yes Yes No Timeout, in seconds, before a resource lock is released, for an editing session of applications, modules, or data tables.

Sample value: 120
Default value: 120

lockKeepalive
(optional)
Yes Yes No Interval, in seconds, before the client sends a ping to the server, to refresh the lock for an editing session of applications, modules, or data tables.

Sample value: 15
Default value: 15

Category: Workflow
Setting Name flowsettings.json tenantsettings.json DesignerEnv Description Value
maxBuilds
(optional)
Yes Yes No Specify the maximum number of builds allowed per application.

Sample value: 20
Default value: 20

enablePTE
(optional)
No No Yes

Section: flowsettings

Boolean value to indicate if PTE objects are enabled at runtime.

Sample value: true
Default value: false

5.5 Features

The features specified in this section are configured under the features object in the flowsettings.json file or the tenantsettings.json file.

For example,

"features": {
           "nexus": true,
..
}
Important
These features are configured only in the flowsettings.json file and the tenantsettings.json file, and not in the DesignerEnv transaction list.
Category Feature Setting Name Mandatory flowsettings.json tenantsettings.json Description Default Value
Audio enableBulkAudioImport Optional Yes Yes Enable/disable the bulk audio import feature. false
grammarValidation Optional Yes yes If this feature is enabled, Designer will validate invalid grammar files during grammar upload and you can upload only valid grammar files (GRXML or Nuance compiled binary grammar files). false
externalAudioSupport Optional Yes Yes If this feature is enabled, a new audio type, External Audio, is available in the Play Message block. It accepts a single variable that contains a URL to the audio resource. MCP will fetch this resource directly and play it. The only supported value of Play As is Audio URI. There is no automatic language switching for this audio type. false
Nexus nexus Optional Yes Yes Enable/disable the Nexus feature. false
Survey survey Optional Yes Yes Enable/disable the survey feature. true
UI Plugins plugins Optional Yes Yes Plugin configuration details.

(Steps are given below the table.)

{}
plugins Optional Yes Yes Enable or disable the plugin feature. false
Milestone enableImplicitModuleMilestones Optional Yes Yes Enable reporting each Shared Module call as an internal milestone. If disabled, Shared Module calls will not generate a milestone. false
Bots enableDialogFlowCXBot Optional Yes Yes When enabled, Dialogflow CX bot type is added to the bot registry and available for selection in the Bot provider drop-down when you configure a new bot. false

5.6 Adding a UI plugin to Designer

  1. Add the plugins array object in the flowsettings.json file (/ofs/designer/flowsettings.json).
    The plugins object contains all the input properties for the plugin app. This is a required property. Whenever there is a change in this object, refresh the browser for the changes to take effect.
    Example:
    "plugins": [
        {
          "url": "http://genesysexample.com/",
          "displayName": "Nexus PII Management",
          "placement": "messageCollections",
          "id":   "nexuspii",
          "mappings": {
                "prod": {
                    "G1-AUS4": "https://genesysexample.com/admin/ux"
                },
                "staging": {
                    "G1-USW1": "http://genesysexample.com/"
                },
            }
        },
        {
           ...
        }]
  2. Add the csplist array object in the flowsettings.json file (/ofs/designer/flowsettings.json).
    The cspList object contains the URL forms to be allowed by Designer's security policy. This is a required property. Whenever there is a change in this object, re-start the node container for the changes to take effect.
    Example:
    If the URL is http://genesysexample.com/, the cspList would be:
    "cspList": ["*.genexample1.com:*", "*.genexample2.com:*", "*.genexample3.com:*"]
  3. Turn on the plugins and nexus feature flags in the Designer tenantSettings.json file (/ofs/<tenantId>/config/tenantSettings.json).
    This is a required property. Whenever there is a change in this object, log out of Designer and log in again for the changes to take effect.
    Important
    If you want to enable the plugins feature for all tenants, add this feature flag in the flowsettings.json file. The feature is enabled for all the tenants under that bucket.
    Example:
    {
        "features": {
            "plugins": true,
            "nexus": true
    }}
  4. Add the url_<plugin-id> property under the plugins section, in Agent Setup. If there is no plugins section, create one. This section is for the tenant URL override. If the DesignerEnv setting (Transactions/Internal/DesignerEnv) is not provided, the plugin URL from the flowsettings.json file is considered.
    This is an optional property. Whenever there is a change in this object, log out of Designer and log in again for the changes to take effect.
    Example:
    {"url_<plugin-id>" : "https://plugin-genesysexample.com"}

6. Logging

Designer and DAS support console output (stdout) logging. Genesys recommends configuring console output logging to minimize the host IOPs and PVCs consumption by using log volumes. Console output logs can be extracted using log collectors like fluentbit/fluentd and Elasticsearch.

Ensure the below setttings are configured in the respective values.yaml overrides for console logging:

  1. Designer
    designerEnv.envs.DES_FILE_LOGGING_ENABLED = false
  2. DAS
    dasEnv.envs.DAS_FILE_LOGGING_ENABLED = false
    dasEnv.envs.DAS_STDOUT_LOGGING_ENABLE = true

6.1 Log levels

Post deployment, Designer and DAS log levels can be modified as follows:

6.1.1 Designer

  1. Configure the logging setting in the flowsettings override (flowsettings.yaml) - Refer to the 5.4 Post deployment configuration settings reference table section for option descriptions.
  2. Execute the steps in the Flowsettings.json update section (see Designer under 8.8 Blue-Green deployment) for the changes to take effect .

6.1.2 DAS

  1. Configure the dasEnv.envs.DAS_LOG_LEVEL setting in the Helm das-values.yaml file. Refer to section 4.2 DAS deployment settings for setting descriptions.
  2. Execute the steps in the Upgrade non production color section (see DAS under 8.8 Blue-Green deployment). The same DAS version running in production can be used for the upgrade,
  3. Execute the steps in the Cutover section (see DAS under 8.8 Blue-Green deployment).

7. Platform / Configuration Server and GWS settings

This section explains the Configuration Server objects and settings required for Designer.

7.1 Create Roles for Designer

Designer uses roles and access groups to determine permissions associated with the logged-in user. To enable this, you must make these changes in GAX or CME.

Designer supports a number of bundled roles suitable for various levels of users.

  • Designer Developer - Most users fall into this category. These users can create Designer applications, upload audio, and create business controls. They have full access to Designer Analytics.
  • Designer Business User - These users cannot create objects but they can manage them (for example, upload audio, change data tables, and view analytics).
  • Designer Analytics - These users only have access to Designer Analytics.
  • Designer Admin - These users can set up and manage partitions associated with users and Designer objects.
  • Designer Operations - Users with this role have full access to all aspects of the Designer workspace. This includes the Operations menu (normally hidden), where they can perform advanced operational and maintenance tasks.

To create these roles, import the .conf files included in the Designer Deployment package. They are located in the packages/roles/ folder.

In addition, ensure the following for user accounts that need access to Designer:

  • The user must have read permissions on its own Person object.
  • Users must be associated with one or more roles via access groups.
  • The on-Premises user must have at least read access on the user, access group(s), and roles(s).
  • The access groups must have read/write permissions to the Agent Setup folders - Scripts and Transactions.

7.2 Create the DesignerEnv transaction list

Designer requires a transaction list for configuration purposes as described in other sections of this document. To set this up:

  1. Create a transaction list called DesignerEnv.
  2. Import the file configuration/DesignerEnv.conf, located in the Designer Deployment Manifest package.
  3. Edit any values according to the descriptions provided in 5.4 Post deployment configuration settings reference table.
  4. Save the list.
  5. Ensure Designer users have at least read access to the DesignerEnv transaction list.

7.3 Platform settings

The platform settings listed below must be configured if the Designer application is used for voice calls.

Component Config Key Value Description
SIP Switch -> Voip Services -> msml service userdata-map-format sip-headers-encoded Option needs to set to pass JSON data as user data in SIPS.
SIP Switch -> Voip Services -> msml service userdata-map-filter * To allow userdata passing to MSML service.
SIPServer --> TServer divert-on-ringing false RONA is handled by the platform.
agent-no-answer-timeout 12
agent-no-answer-action notready
agent-no-answeroverflow "" No value, empty.
after-routing-timeout 24
sip-treatments-continuous true
msml-record-support true To allow routed calls recording via the Media Server.
Switch object annex --> gts ring-divert 1
ORS --> orchestration new-session-on-reroute false Required for SIPS Default Routing (Default Routing handling (Voice)).
MCP [vxmli] transfer.allowed TRUE Required for Transfer block (allows VXML Transfer in MCP).
MCP [cpa] outbound.method NATIVE Required for Transfer block (allow CPA detection for Transfer ).
UCS [cview] enabled TRUE Enables Customer Context Services.

7.4 GWS configuration

Ensure that the following steps are performed in GWS.

7.4.1 Create Contact Center

Create a contact center in GWS if it is not already created. Refer to the GWS documentation for more information on this.

7.4.2 Create GWS Client

Create new GWS client credentials if they are not already created . Refer to the GWS documentation for more information on this.

8. Deployment

This section describes the deployment process for Designer and DAS.

8.1 Preparation

Before you deploy Designer and DAS using Helm charts, complete the following preparatory steps:

  1. Ensure the Helm client is installed.
  2. Set up an Ingress Controller, if not already done.
  3. Setup an NFS server, if not already done.
  4. Create Persistent Volumes - a sample YAML file is provided in the Designer manifest package.
  5. Download the Designer and DAS docker images and push to the local docker registry.
  6. Download the Designer package and extract to the current working directory.
  7. Configure Designer and DAS value overrides (designer-values.yaml and das-values.yaml); ensure the mandatory settings are configured. If the Blue-Green deployment process is used, Ingress settings are explained in the 8.8 Blue-Green deployment section.

8.2 Set up Ingress

Given below are the requirements to set up an Ingress for the Designer UI:

  • Cookie name - designer.session.
  • Header requirements - client IP & redirect, passthrough.
  • Session stickiness - enabled.
  • Allowlisting - optional.
  • TLS for ingress - optional (should be able to enable or disable TLS on the connection).

8.3 Set up Application Gateway (WAF) for Designer

Designer Ingress must be exposed to the internet using Application Gateway enabled with WAF.

When WAF is enabled, consider the following exception in the WAF rules for Designer:

  • Designer sends a JSON payload with data, for example, {profile . {} }. Sometimes, this is detected as OSFileAccessAttempt, which is a false positive detection. Disable this rule if you encounter a similar issue in your WAF setup.

8.4 Storage

8.4.1 Designer storage

Designer requires storage to store designer application workspaces. Designer storage is a shared file storage that will be used by the Designer and DAS services.
Important
This storage is critical. Ensure you take backups and snapshots at a regular interval, probably, each day.
A Zone-Redundant Storage system is required to replicate data from the RWX volumes and must be shared across multiple pods:
  • Capacity - 1 TiB
  • Tier - Premium
  • Baseline IO/s - 1424
  • Burst IO/s - 4000
  • Egress Rate - 121.4 MiBytes/s
  • Ingress Rate - 81.0 MiBytes/s

8.4.2 Permission considerations for Designer and DAS storage

NFS

For NFS RWX storages, the mount path should be owned by <code>genesys:genesys</code>, that is, <code>500:500</code> with <code>0777</code> permissions. It can be achieved by one of the below methods:

  • From the NFS server, execute the chmod -R 777 <export_path> and chown -R 500:500 <export_path> commands to set the required permissions.
  • Create a dummy Linux based pod that mounts the NFS storage. From the pod, execute the chmod -R 777 <mount_path> and chown -R 500:500 <mount_path> commands. This sets the required permissions. However, this method might require the Linux based pods to be run as privileged.

SMB / CIFS

For SMB / CIFS based RWX storages, for instance, Azure file share, the below <code>mountOptions</code> must be used in the StorageClass or the PersistentVolume template:

<code>mountOptions</code>
- dir_mode=0777
- file_mode=0777
- uid=500
- gid=500
- mfsymlinks
- cache=strict

8.5 Set up Secrets

Secrets are required by the Designer service to connect to GWS and Redis (if you are using them).

GWS Secrets:

  • GWS provides a Client ID and secrets to all clients that can be connected. You can create Secrets for the Designer client as specified in the Set up secrets for Designer section below.

Redis password:

  • If Designer is connected to Redis, you must provide the Redis password to Designer to authenticate the connection.

8.5.1 Set up Secrets for Designer

Use the designer.designerSecrets parameter in the values.yaml file and configure Secrets as follows:

designerSecrets:
  enabled: true
  secrets:
    DES_GWS_CLIENT_ID: xxxx
    DES_GWS_CLIENT_SECRET: xxxx
    DES_REDIS_PASSWORD: xxxxx

8.6 Deployment strategies

Designer supports the following deployment strategies:

  • Rolling Update (default).
  • Blue-Green (recommended).

DAS (Designer Application Server) supports the following deployment strategies:

  • Rolling Update (default).
  • Blue-Green (recommended).
  • Canary (must be used along with Blue-Green and is recommended in production).

8.7 Rolling Update deployment

The rolling update deployment is the standard default deployment to Kubernetes. It works slowly, one by one, replacing pods of the previous version of your application with pods of the new version without any cluster downtime. It is the default mechanism of upgrading for both Designer and DAS.

8.7.1 Designer

Initial deployment

To perform the initial deployment for a rolling upgrade in Designer, use the Helm command given below. The values.yaml file can be created as required.

  • helm upgrade --install designer -f designer-values.yaml designer-100.0.112+xxxx.tgz --set designer.image.tag=9.0.1xx.xx.xx

The values.yaml overrides passed as an argument to the above Helm upgrade command:

designer.image.tag=9.0.1xx.xx.xx - This is the new Designer version to be installed, for example, 9.0.111.05.5.

Upgrade

To perform an upgrade, the image version has to be upgraded in the designer-values.yaml file or can be set using the --set flag through the command given below. Once the designer-values.yaml file is updated, use this Helm command to perform the upgrade:

  • helm upgrade --install designer -f designer-values.yaml designer-100.0.112+xxxx.tgz --set designer.image.tag=9.0.1xx.xx.xx

The values.yaml overrides passed as an argument to the above Helm upgrade command:

designer.image.tag=9.0.1xx.xx.xx - This is the new Designer version to be installed, for example, 9.0.111.05.5.

Rollback

To perform a rollback, the image version in the designer-values.yaml file can be downgraded. Or you can use the --set flag through the command given below. Once the designer-values.yaml file is updated, use this Helm command to perform the rollback:

  • helm upgrade --install designer -f designer-values.yaml designer-100.0.112+xxxx.tgz --set designer.image.tag=9.0.1xx.xx.xx

The values.yaml overrides passed as an argument to the above Helm upgrade command:

designer.image.tag=9.0.1xx.xx.xx - This is the Designer version to be rolled back to, for example, 9.0.111.05.5.

8.7.2 DAS

Initial deployment

To perform the initial deployment for a rolling upgrade in DAS, use the Helm command given below. The values.yaml file can be created as required.

  • helm upgrade --install designer-das -f designer-das-values.yaml designer-das-100.0.112+xxxx.tgz --set das.image.tag=9.0.1xx.xx.xx

The values.yaml overrides passed as an argument to the above Helm upgrade command:

das.image.tag=9.0.1xx.xx.xx - This is the new DAS version to be installed, for example, 9.0.111.05.5.

Upgrade

To perform an upgrade, the image version has to be upgraded in the designer-das-values.yaml file or can be set using the --set flag through the command given below. Once the designer-das-values.yaml file is updated, use this Helm command to perform the upgrade:

  • helm upgrade --install designer-das -f designer-das-values.yaml designer-das-100.0.112+xxxx.tgz --set das.image.tag=9.0.1xx.xx.xx

The values.yaml overrides passed as an argument to the above Helm upgrade command:

das.image.tag=9.0.1xx.xx.xx - This is the new DAS version to be installed, for example, 9.0.111.05.5.

Rollback

To perform a rollback, the image version in the designer-das-values.yaml file can be downgraded. Or you can use the --set flag through the command given below. Once the designer-das-values.yaml file is updated, use this Helm command to perform the rollback:

  • helm upgrade --install designer-das -f designer-das-values.yaml designer-das-100.0.112+xxxx.tgz --set das.image.tag=9.0.1xx.xx.xx

The values.yaml overrides passed as an argument to the above Helm upgrade command:

das.image.tag=9.0.1xx.xx.xx - This is the DAS version to be rolled back to, for example, 9.0.111.05.5.

8.8 Blue-Green deployment

Blue-Green deployment is a release management technique that reduces risk and minimizes downtime. It uses two production environments, known as Blue and Green or active and inactive, to provide reliable testing, continuous no-outage upgrades, and instant rollbacks.When a new release needs to be rolled out, an identical deployment of the application will be created using the Helm package and after testing is completed, the traffic is moved to the newly created deployment which becomes the active environment, and the old environment becomes inactive. This ensures that a fast rollback is possible by just changing route if a new issue is found with live traffic. The old inactive deployment is removed once the new active deployment becomes stable.

8.8.1 Designer

Service cutover is done by updating the Ingress rules. The diagram below shows the high-level approach to how traffic can be routed to Blue and Green deployments with Ingress rules.

DesBlueGreenDep.png

Preparation

Before you deploy Designer using the blue-green deployment strategy, complete the following preparatory steps:

  1. Create 3 hostnames as given below. The blue service hostname must contain the string blue. For example, designer.blue.example.com or designer-blue.example.com. The green service hostname must contain the string green. For example, designer.green.example.com or designer-green.example.com. The blue/green services can be accessed separately with the blue/green hostnames:
    • designer.example.com - For the production host URL, this is used for external access.
    • designer.blue.example.com - For the blue service testing.
    • designer.green.example.com - For the green service testing.
  2. Configure the hostnames in the designer-values.yaml file under ingress. Annotations and paths can be modified as required.
    ingress:
      enabled: true
      annotations: {}
      paths: [/]
      hosts:
        - designer.example.com
        - designer.blue.example.com
        - designer.green.example.com

Initial deployment

The resources - ingress and persistent volume claims (PVC) - must be created initially before deploying the Designer service as these resources are shared between blue/green services and they are required to be created at the very beginning of the deployment. These resources are not required for subsequent upgrades. The required values are passed using the -- set flag in the following steps. Values can also be directly changed in the values.yaml file.

  1. Create Persistent Volume Claims required for the Designer service (assuming the volume service name is designer-volume).
    helm upgrade --install designer-volume -f designer-values.yaml designer-9.0.xx.tgz --set designer.deployment.strategy=blue-green-volume
    The values.yaml overrides passed as an argument to the above Helm upgrade command:
    designer.deployment.strategy=blue-green-volume - This denotes that the Helm install will create a persistent volume claim in the blue/green strategy.
  2. Create Ingress rules for the Designer service (assuming the ingress service name will be designer-ingress):
    helm upgrade --install designer-ingress -f designer-values.yaml designer-100.0.112+xxxx.tgz --set designer.deployment.strategy=blue-green-ingress --set designer.deployment.color=green
    The values.yaml overrides passed as an argument to the above Helm upgrade command:
    designer.deployment.strategy=blue-green-ingress - This denotes that the Helm install will create ingress rules for the Designer service.
    designer.deployment.color=green - This denotes that the current production (active) color is green.
  3. Deploy the Designer service color selected in step 2. In this case, green is selected and assuming the service name is designer-green:
    helm upgrade --install designer-green -f designer-values.yaml designer-100.0.112+xxxx.tgz --set designer.deployment.strategy=blue-green --set designer.image.tag=9.0.1xx.xx.xx --set designer.deployment.color=green

Upgrade

  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.
    DesUpgStep1.png
  2. Deploy the Designer service on to the non-production color. In the above example, blue is the non-production color and assuming the service name will be designer-blue:
    helm upgrade --install designer-blue -f designer-values.yaml designer-100.0.112+xxxx.tgz --set designer.deployment.strategy=blue-green --set designer.image.tag=9.0.1xx.xx.xx --set designer.deployment.color=blue
    The values.yaml overrides passed as an argument to the above Helm upgrade command:
    designer.deployment.strategy=blue-green - This denotes that the Designer service is installed using the blue-green strategy.
    designer.image.tag=9.0.1xx.xx.xx - This denotes the new Designer version to be installed, for example, 9.0.116.08.12.
    designer.deployment.color=blue - This denotes that the blue color service is installed.
    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+xxxx.tgz --set designer.deployment.strategy=blue-green --set designer.image.tag=9.0.1xx.xx.xx --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:

  1. 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+xxxx.tgz --set designer.deployment.strategy=blue-green-ingress --set designer.deployment.color=blue
    The values.yaml overrides passed as an argument to the above Helm upgrade command:
    designer.deployment.strategy=blue-green-ingress - This denotes that the helm install will create ingress rules for the Designer service.
    designer.deployment.color=blue - This denotes that the current production (active) color is blue.
  2. 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.

Rollback

If the upgrade must be rolled back, the ingress rules can be modified to point to the old deployment pods (green, in this example) by performing a cutover again.

  1. Perform a cutover using the following command:
    helm upgrade --install designer-ingress -f designer-values.yaml designer-100.0.112+xxxx.tgz --set designer.deployment.strategy=blue-green-ingress --set designer.deployment.color=green
    The values.yaml overrides passed as an argument to the above Helm upgrade command:
    designer.deployment.strategy=blue-green-ingress - This denotes that the Helm install will create Ingress rules for the Designer service.
    designer.deployment.color=green - This denotes that the the current production (active) color is green.
  2. Verify the Ingress rules by running the following command:
    kubectl describe ingress designer-ingress
    The production host name must point to the green service.

8.8.2 DAS

As with Designer, the Blue-Green strategy can be adopted for DAS as well. The Blue-Green architecture used for DAS is given below. Here, the cutover mechanism is controlled by Service, the Kubernetes manifest responsible for exposing the pods. The Ingress, when enabled, will point to the appropriate service based on the URL.

DASBlueGreenDep.png

Ingress setup

Important
Ingress for DAS must be enabled only if DAS has to be reached from outside the Kubernetes cluster. If you don't intend to expose DAS outside the cluster, then ingress need not be enabled and you can skip these steps.
  1. Configure Ingress Host names for DAS.
    Create 3 hostnames as follows: The blue service host name must contain the string blue; for instance, das.blue.example.com or das-blue.example.com, The green service host name must contain the string green; for instance, das.green.example.com or das-green.example.com. The green/blue services can be accessed separately with these blue/green hostnames.
    * das.example.com - This is the production host url, and is used for external access.
    * das.blue.example.com - This is for blue service testing.
    * das.green.example.com - This is for green service testing.
  2. Configure the hostnames in the das-values.yaml file under ingress, annotations, and paths (can be modified based on the requirement).
    ingress:
      enabled: true
      annotations: {}
      paths: ["/"]
      hosts:
        - das.example.com
        - das.blue.example.com
        - das.green.example.com

Initial deployment

The Ingress must be created initially before deploying the DAS service since it is shared between blue/green services and it is required to be created at the very beginning of the deployment. The Ingress is not required for subsequent upgrades. The required values are passed using the -- set flag in the following steps. Values can also be directly changed in the values.yaml file.

  1. Deploy initial DAS pods and other resources by choosing an active color, in this example, green. Use the below command to create a designer-das-green service:
    helm upgrade --install designer-das-green -f designer-das-values.yaml designer-das-100.0.106+xxxx.tgz --set das.deployment.strategy=blue-green --set das.image.tag=9.0.1xx.xx.xx --set das.deployment.color=green
    The values.yaml overrides passed as an argument to the above Helm upgrade command:
    das.deployment.strategy=blue-green - This denotes that the DAS service will be installed using the blue-green deployment strategy.
    das.image.tag=9.0.1xx.xx.xx - This denotes the DAS version to be installed, for example, 9.0.111.04.4.
    das.deployment.color=green - This denotes that the green color service is installed.
  2. Once the initial deployment is done, the pods have to be exposed to the <code>designer-das</code> service. Execute the following command to create the <code>designer-das</code> service:
    helm upgrade --install designer-das designer-das-100.0.106+xxx.tgz -f designer-das-values.yaml --set das.deployment.strategy=blue-green-service --set das.deployment.color=green
    The values.yaml overrides passed as an argument to the above helm upgrade
    das.deployment.strategy=blue-green-service - This denotes that the designer-das service will be installed and exposed to the active color pods.
    das.deployment.color=green - This denotes that the designer-das service will point to green pods.
  3. Create ingress rules for the DAS service (assuming the ingress service is das-ingress):
helm upgrade --install das-ingress designer-das-100.0.106+xxxx.tgz -f designer-das-values.yaml --set das.deployment.strategy=blue-green-ingress

The values.yaml overrides passed as an argument to the above command
das.deployment.strategy=blue-green-ingress - This denotes that the helm install will create ingress rules for the DAS service.

Important
Step 3 is required only when ingress is to be created to expose DAS outside the cluster.

NodePort Service

The designer-das-green release creates a service called designer-das-green and the designer-das-blue release creates a service called designer-das-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 30280. If this was applied to designer-das-green, use a different value for designer-das-blue, for example, 30281. Use the below helm command to achieve this: helm upgrade --install designer-das designer-das-100.0.106+xxx.tgz -f designer-das-values.yaml --set das.deployment.strategy=blue-green-service --set das.deployment.color=green --set das.service.nodePort=30281

8.9 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. If you are not opting for Canary, skip the steps in this section.

Canary deployment

  1. Identify the current production color by checking the designer-das service selector labels (kubectl describe service designer-das). Green is the production color in the below example as the selector label is color=green.
    CanaryInDepStep1.png
  2. To deploy canary pods, the das.deployment.strategy value must be set to canary in the designer-das-values.yaml file or using the -- set flag as shown in the command below:
    helm upgrade --install designer-das-canary -f das-values.yaml designer-das-100.0.106+xxxx.tgz --set das.deployment.strategy=canary --set das.image.tag=9.0.1xx.xx.xx --set das.deployment.color=green
    The values.yaml overrides passed as an argument to the above Helm upgrade command:
    das.deployment.strategy=canary - This denotes that the Helm install will create canary pods.
    das.deployment.color=green - This denotes that the current production (active) color is green.
    Important
    To make sure Canary pods receive live traffic, they have to be exposed to the designer-das service by setting das.deployment.color=<active_color>, which is obtained from step 1.
  3. Once canary pods are up and running, ensure that the designer-das service points to the canary pods using the kubectl describe svc designer-das command.
    CanaryEndpoints.png
    The IP address present in the Endpoints must match the IP address of the canary pod. The canary pod's IP address is obtained using the kubectl describe pod <canary_pod_name> command.
    CanaryIPs.png

Cleaning up

After completing canary testing, the canary pods must be cleaned up.

The das.deployment.replicaCount must be made zero and the release is upgraded. It can be changed in the designer-das-values.yaml file or through the --set flag as follows:

  • helm upgrade --install designer-das-canary -f das-values.yaml designer-das-100.0.106+xxxx.tgz --set das.deployment.strategy=canary --set das.image.tag=9.0.1xx.xx.xx --set das.deployment.color=blue --set das.deployment.replicaCount=0

Upgrade

  1. Identify the current production color by checking the designer-das service selector labels (kubectl describe service designer-das). Green is the production color in the below example as the selector label is color=green.
    DasBGUpgStep1.png
  2. Deploy the DAS service on to the non-production color. For the above 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+xxxx.tgz --set das.deployment.strategy=blue-green --set das.image.tag=9.0.1xx.xx.xx --set das.deployment.color=blue
    The values.yaml overrides passed as an argument to the above Helm upgrade command:
    das.deployment.strategy=blue-green - This denotes that the DAS service is installed using the blue-green strategy.
    das.image.tag=9.0.1xx.xx.xx - This denotes the new DAS version to be installed, for example, 9.0.111.05.5.
    das.deployment.color=blue - This denotes that the blue color service is installed.
    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.

  1. Update the designer-das service with the new deployment color by executing the below command. In this example, blue is the new deployment color (non-production color).
    helm upgrade --install designer-das-service -f designer-das-values.yaml designer-das-100.0.106+xxxx.tgz --set das.deployment.strategy=blue-green-service --set das.deployment.color=blue
  2. 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.

Rollback

  1. If the upgrade must be rolled back, cutover has to performed again to make the service point to the old deployment (green) again. Use the below command to perform the cutover:
    helm upgrade --install designer-das-service -f designer-das-values.yaml designer-das-100.0.106+xxxx.tgz --set das.deployment.strategy=blue-green-service --set das.deployment.color=green
    The values.yaml overrides passed as an argument to the above Helm upgrade command:
    das.deployment.strategy=blue-green-service - This denotes that the Helm install will create ingress rules for the DAS service.
    das.deployment.color=green - This denotes that the current production (active) color is green.
  2. Verify the service by executing the kubectl describe service designer-das the command. The type label must have the active color's label, that is, color=green.

8.10 Validations and checks

Here are some common validations and checks that can be performed to know if the deployment was successful.

  • Check if the application pods are in running state by using the kubectl get pods command.
  • Try to connect to the Designer or DAS URL as per the ingress rules from your browser. You must be able to access the Designer and DAS webpages.


9. Post deployment procedures

Upgrading the Designer workspace

Workspace resources must be upgraded after cutover. This will 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 will create 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 will upgrade 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)).
Important
The above steps will also be used for further upgrades.

Updating the flowsettings file

Post deployment, the flowsettings.json file can be modified through a Helm install as follows:

  1. Extract the Designer Helm Chart and find the flowsettings.yaml file under the Designer Chart > Config folder.
  2. Modify the necessary settings (refer to the Post deployment configuration settings reference table for the different settings and their allowed values).
  3. Execute the below Helm upgrade command on the non-production color service. It can be done as part of the Designer upgrade by passing the flowsettings.yaml file using the --values flag. In this case, a new Designer version can be used for the upgrade. If it is only a flowsettings.json update, the same Designer version is used.
    helm upgrade --install designer-blue -f designer-values.yaml -f flowsettings.yaml designer-9.0.xx.tgz --set designer.deployment.strategy=blue-green --set designer.image.tag=9.0.1xx.xx.xx --set designer.deployment.color=blue
  4. Once testing is completed on the non-production service, perform the cutover step as mentioned in the Cutover section (Designer Blue-Green deployment). After cutover, the production service will contain the updated settings. The non-active color Designer must also be updated with the updated settings after the cutover.

10. Enabling optional features

10.1 Enable Designer Analytics and Audit Trail

Post Designer deployment, features such as Analytics and Audit Trail can be enabled by performing the below steps.

Important
Ensure Elasticsearch is deployed before proceeding.

10.1.1 Designer changes

  1. Configure the following settings in flowsettings override (flowsettings.yaml) - Refer to the 5.4 Post deployment configuration settings reference table section for option descriptions.
    • enableAnalytics: true
    • enableESAuditLogs: true
    • esServer
    • esPort
    • esUrl
  2. Configure the below setting in the DesignerEnv transaction list:
    ReportingURL in the reporting section.
  3. Perform the steps in the Updating the flowsettings file section under 9. Post deployment procedures.

10.1.2 DAS changes

  1. Configure the following settings in the helm das-values.yaml file. Refer to the 4.2 DAS deployment settings section for setting descriptions.
    dasEnv.envs.DAS_SERVICES_ELASTICSEARCH_ENABLED = true
    dasEnv.envs.DAS_SERVICES_ELASTICSEARCH_HOST
    dasEnv.envs.DAS_SERVICES_ELASTICSEARCH_PORT
  2. Perform the steps in the Upgrade non production color section (see DAS under 8.8 Blue-Green deployment). The same DAS version running in production can be used for the upgrade.
  3. Perform the steps in the Cutover section (see DAS under 8.8 Blue-Green deployment).

10.2 Enable Personas

You can enable the Personas feature in Designer by following the below steps.

10.2.1 Deploy personas.json

  • Deploy the personas.json file in the workspace location, /workspace/{tenantID}/workspace/personas/personas.json.
  • Create the personas directory if it does not exist.
Given below is a sample personas.json file:
[
    {
        "id": "1",
        "name": "Samantha",
        "gender": "female",
        "tags": ["female", "middle-age", "default"],
        "displayPersona": "female, 30-40s, professional, calm",
        "voice": [{
                "name": "samantha",
                "language": "en-US",
                "ttsname": "Samantha",
                "ttsengine": "NuanceTTS",
                "displayName": "Samantha"
            }, {
                "name": "karen",
                "language": "en-AU",
                "ttsname": "Karen",
                "ttsengine": "NuanceTTS",
                "displayName": "Karen"
            }, {
                "name": "amelie",
                "language": "fr-CA",
                "ttsname": "Amelie",
                "ttsengine": "NuanceTTS",
                "displayName": "Amelie"
            }, {
                "name": "paulina",
                "language": "es-MX",
                "ttsname": "Paulina",
                "ttsengine": "NuanceTTS",
                "displayName": "Paulina"
            }
        ],
        "digital": {},
        "email": {},
        "chat": {},
        "web": {}
    },
    {
        "id": "2",
        "name": "Tom",
        "gender": "male",
        "tags": ["male", "middle-age"],
        "displayPersona": "male, 30-40s, polite, professional",
        "voice": [{
                "name": "tom",
                "language": "en-US",
                "ttsname": "Tom",
                "ttsengine": "NuanceTTS",
                "displayName": "Tom"
            }, {
                "name": "lee",
                "language": "en-AU",
                "ttsname": "Lee",
                "ttsengine": "NuanceTTS",
                "displayName": "Lee"
            }, {
                "name": "felix",
                "language": "fr-CA",
                "ttsname": "Felix",
                "ttsengine": "NuanceTTS",
                "displayName": "Felix"
            }, {
                "name": "javier",
                "language": "es-MX",
                "ttsname": "Javier",
                "ttsengine": "NuanceTTS",
                "displayName": "Javier"
            }
        ],
        "digital": {},
        "email": {},
        "chat": {},
        "web": {}
    },
    {
        "id": "3",
        "name": "Gabriela",
        "gender": "female",
        "tags": ["female", "young", "engaging"],
        "displayPersona": "female, 20-30s, engaging",
        "voice": [{
                "name": "gabriela",
                "language": "en-US",
                "ttsname": "en-US-Standard-E",
                "ttsengine": "GTTS",
                "displayName": "Gabriela"
            }, {
                "name": "sheila",
                "language": "en-AU",
                "ttsname": "en-AU-Standard-A",
                "ttsengine": "GTTS",
                "displayName": "Sheila"
            }, {
                "name": "lili",
                "language": "fr-CA",
                "ttsname": "fr-CA-Standard-A",
                "ttsengine": "GTTS",
                "displayName": "Lili"
            }
        ],
        "digital": {},
        "email": {},
        "chat": {},
        "web": {}
    },
    {
        "id": "4",
        "name": "Michael",
        "gender": "male",
        "tags": ["male", "young"],
        "displayPersona": "male, 20-30s, curious, geeky",
        "voice": [{
                "name": "michael",
                "language": "en-US",
                "ttsname": "en-US-Standard-B",
                "ttsengine": "GTTS",
                "displayName": "Michael"
            }, {
                "name": "royce",
                "language": "en-AU",
                "ttsname": "en-AU-Standard-B",
                "ttsengine": "GTTS",
                "displayName": "Royce"
            }, {
                "name": "alexandre",
                "language": "fr-CA",
                "ttsname": "fr-CA-Standard-B",
                "ttsengine": "GTTS",
                "displayName": "Alexandre"
            }
        ],
        "digital": {},
        "email": {},
        "chat": {},
        "web": {}
    },
    {
        "id": "5",
        "name": "Diane",
        "gender": "female",
        "tags": ["female", "mature"],
        "displayPersona": "female, 40-50s, soothing, silky",
        "voice": [{
                "name": "diane",
                "language": "en-US",
                "ttsname": "en-US-Standard-C",
                "ttsengine": "GTTS",
                "displayName": "Diane"
            }, {
                "name": "muriel",
                "language": "en-AU",
                "ttsname": "en-AU-Standard-C",
                "ttsengine": "GTTS",
                "displayName": "Muriel"
            }, {
                "name": "chloe",
                "language": "fr-CA",
                "ttsname": "fr-CA-Standard-C",
                "ttsengine": "GTTS",
                "displayName": "Chloe"
            }
        ],
        "digital": {},
        "email": {},
        "chat": {},
        "web": {}
    },
    {
        "id": "6",
        "name": "David",
        "gender": "male",
        "tags": ["male", "mature"],
        "displayPersona": "male, 40-50s, professional, confident",
        "voice": [{
                "name": "david",
                "language": "en-US",
                "ttsname": "en-US-Standard-D",
                "ttsengine": "GTTS",
                "displayName": "David"
            }, {
                "name": "austin",
                "language": "en-AU",
                "ttsname": "en-AU-Standard-D",
                "ttsengine": "GTTS",
                "displayName": "Austin"
            }, {
                "name": "pierre",
                "language": "fr-CA",
                "ttsname": "fr-CA-Standard-D",
                "ttsengine": "GTTS",
                "displayName": "Pierre"
            }
        ],
        "digital": {},
        "email": {},
        "chat": {},
        "web": {}
    }
]

10.2.2 Update Designer flowsettings.json

  • Enable the persona feature flag in the flowsettings.json override file.
"features": {
  "persona": true

Update application settings

Perform the following steps to enable the persona in the required Designer application:

  1. Open the required Designer application and navigate to the Settings tab.
  2. In Application Settings, select the Enable Persona checkbox in the Persona tab.
  3. Re-publish the application and create a new build.

11. Cleanup

11.1 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-*

12. Limitations

Designer currently supports multi-tenancy provided by the tenant Configuration Server. That is, each tenant should have a dedicated Configuration Server, and Designer can be shared across the multiple tenants.

Comments or questions about this documentation? Contact us for support!