GraphRAG Installation

###############################################################################
# Main configuration file                                                     #
# To override single property use --set                                       #
# To override multiple, provide another values-override.yaml with the -f flag #
# See https://helm.sh/docs/chart_template_guide/values_files/                 #
###############################################################################

#########################
# Global Configurations #
#########################

global:
  # Overrides image.registry
  # Can be used to override it globally when using umbrella charts.
  imageRegistry: ""

  # Inserts additional pull secret references together along with any from .Values.image.pullSecrets
  # Can be used to override it globally when using umbrella charts.
  # Values are processed as Helm templates.
  # Ref: https://kubernetes.io/docs/concepts/configuration/secret/#using-imagepullsecrets
  #
  # Example:
  # imagePullSecrets:
  #   - name: my-pull-secret
  imagePullSecrets: []

###############################
#   Metadata Configurations   #
# Naming and labels overrides #
###############################

# Overrides the name of the chart affecting the names of the GraphRAG Chatbot resources.
nameOverride: ""

# Overrides the naming of all GraphRAG Chatbot resources, effectively removing the chart's name and the release name prefix.
# This override takes precedence over anything in .Values.nameOverride
fullnameOverride: ""

# Overrides the deployment namespace in case of multi-namespace deployments, for example when using umbrella charts where some sub-charts should be
# deployed in different namespaces.
# This affects every resource deployed by this chart.
# The default value is .Release.Namespace if this is left unspecified.
namespaceOverride: ""

# Additional common labels to add to all GraphRAG Chatbot resources.
# Values are processed as Helm templates.
# Ref: https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/
#
# Example:
# labels:
#   foo: bar
#   some-label: {{ .Values.someValue }}
labels: {}

# Additional common annotations to add to all GraphRAG Chatbot resources.
# Values are processed as Helm templates.
# Ref: https://kubernetes.io/docs/concepts/overview/working-with-objects/annotations/
#
# Example:
# annotations:
#   foo: bar
#   some-annotation: {{ .Values.someValue }}
annotations: {}

#########################################
# GraphRAG Chatbot Image Configurations #
#########################################

# Image configurations for GraphRAG Chatbot service.
# Ref: https://kubernetes.io/docs/concepts/containers/images/
image:
  # The registry that hosts the GraphRAG Chatbot image.
  # The default is to pull it from the official Docker registry, but this can be overridden to pull from other public or private registries.
  registry: maven.ontotext.com

  # The repository name of the GraphRAG Chatbot image.
  repository: graphrag-ui/chatbot

  # Image tag that corresponds to the version of GraphRAG Chatbot.
  # By default, the chart uses .Chart.AppVersion to construct the full image name.
  # Use this to override the value from .Chart.AppVersion and effectively deploy a custom GraphRAG Chatbot version.
  tag: ""

  # Expected SHA256 digest of the used GraphRAG Chatbot image, e.g. "sha256@abc"
  # Use the digest to make sure you are always deploying the exact same GraphRAG Chatbot image.
  # Defining this would override .Chart.AppVersion and image.tag
  digest: ""

  # Defines the policy for pulling images
  # Ref: https://kubernetes.io/docs/concepts/containers/images/#image-pull-policy
  pullPolicy: IfNotPresent

  # Secrets for pulling GraphRAG Chatbot's image from secured registries.
  # Values are processed as Helm templates.
  # Ref: https://kubernetes.io/docs/concepts/configuration/secret/#using-imagepullsecrets
  #
  # Example:
  # pullSecrets:
  #   - name: my-pull-secret
  pullSecrets: []

##########################
# Scaling Configurations #
##########################

# Number of GraphRAG Chatbot pods to be deployed as part of the Deployment resources.
replicas: 1

# Horizontal Pod Autoscaler configuration that can automatically scale the number of GraphRAG Chatbot pods based on
# resource utilization.
# Ref: https://kubernetes.io/docs/concepts/workloads/autoscaling/
autoscaling:
  enabled: false
  minReplicas: 1
  maxReplicas: 10
  targetCPUUtilizationPercentage: 80
  targetMemoryUtilizationPercentage: 80

###################################
# GraphRAG Chatbot Configurations #
###################################

# GraphRAG Chatbot runtime configuration settings.
# For reference, see https://gitlab.ontotext.com/graphwise-graphrag/graphrag-ui
configuration:

  # The external URL at which GraphRAG Chatbot would be accessible.
  # This can be a publicly available domain name, an internal one or even a Kubernetes service address.
  #
  # It also configures the hostname in the default Ingress resource, if enabled.
  # The value is processed as a Helm template.
  #
  # Note: If the external URL uses HTTPS and the default Ingress is enabled, you have to configure .Values.ingress.tls
  #
  # Note: When deploying on a context path different from /, you need to properly configure the Ingress according to the requirements of the
  # Ingress controller implementation.
  externalUrl: http://graphrag.127.0.0.1.nip.io/

  # Configurations for inserting an additional NGINX configuration snippet into the default.conf file that will proxy pass requests to the
  # GraphRAG Conversation API via the chatbot NGINX.
  #
  # By default, the proxy pass is enabled. so you don't need to expose the Conversation API separately.
  # You just need to set GRAPHRAG_BACKEND_URL with a context path.
  conversationProxy:
    # Toggles the proxy pass configuration snippet in the default NGINX configuration.
    enabled: true

    # The service URL where the GraphRAG Conversation API is accessible.
    #
    # This is used to configure proxy locations in the NGINX configuration that will redirect requests to the GraphRAG Conversation API.
    # Ideally, this should be the internal service address of the GraphRAG Conversation API.
    #
    # The value is processed as a Helm template.
    url: http://conversation:8080/

  # Main configurations for the GraphRAG Chatbot web application.
  # These are environment variables that are passed to the GraphRAG Chatbot Nginx which then uses them to configure different URLs.
  properties:
    GRAPHRAG_BACKEND_URL: "http://graphrag.127.0.0.1.nip.io/api"
    GRAPHRAG_KEYCLOAK_URL: "http://auth.127.0.0.1.nip.io/"
    GRAPHRAG_KEYCLOAK_REALM: "graphrag"
    GRAPHRAG_KEYCLOAK_CLIENT_ID: "graphrag-web"

  # NGINX configurations for the GraphRAG Chatbot web application.
  nginx:
    # Reference to an existing ConfigMap resource containing a default.conf NGINX configuration file for the GraphRAG Chatbot web application.
    # Providing this overrides the default.conf file that comes with the chart.
    #
    # The value is processed as a Helm template.
    defaultConfConfigmap: ""
    # The key in the ConfigMap resource that contains the default.conf NGINX configuration file.
    defaultConfConfigmapKey: "default.conf"

##########################
# Ingress Configurations #
##########################

# Configurations for the default Ingress resource for GraphRAG Chatbot.
#
# The default Ingress in this chart makes no assumptions of what Ingress controller is being used, it's up to you to define any specific annotations
# required by the controller implementation in your cluster.
#
# Ref: https://kubernetes.io/docs/concepts/services-networking/ingress/
ingress:
  # Toggles the deployment of the default Ingress resource.
  enabled: true
  # Specifies the ingress controller implementation that will control this Ingress resource.
  # Not defining this would result in using the default ingress controller in the cluster, if there is one.
  className: ""
  # Additional labels to append to the Ingress resource.
  # Values are processed as Helm templates.
  labels: {}
  # Additional annotations to append to the Ingress resource.
  # Values are processed as Helm templates.
  annotations: {}
  # If set, overrides the host from .Values.configuration.externalUrl
  host: ""
  # If set, overrides the context path from .Values.configuration.externalUrl
  path: ""
  # Sets the ingress path type.
  # If you need to use ImplementationSpecific, make sure to set any annotations needed by the controller implementation.
  pathType: Prefix
  # Configures SSL termination on Ingress level.
  # Ref: https://kubernetes.github.io/ingress-nginx/examples/tls-termination/
  tls:
    # Feature toggle for SSL termination. Disabled by default.
    # If TLS is enabled, the .Values.configuration.externalUrl should also be updated to use HTTPS.
    enabled: false
    # Name of a Kubernetes Secret object with the key and certificate.
    # If TLS is enabled, it's required to be provided, depending on the deployment.
    # This could be an existing Secret or one that is not yet created.
    secretName: ""
  # List of additional hostnames to append to the Ingress resource.
  # Values are processed as Helm templates.
  extraHosts: []
  # List of additional TLS records to append to the Ingress resource.
  # Values are processed as Helm templates.
  extraTLS: []

##########################
# Service Configurations #
##########################

# Configurations for GraphRAG Chatbot Service.
# Ref: https://kubernetes.io/docs/concepts/services-networking/service/
service:
  # Enables or disables the Service deployment
  enabled: true
  # Additional labels to append to the Service resource.
  # Values are processed as Helm templates.
  labels: {}
  # Additional annotations to append to the Service resource.
  # Values are processed as Helm templates.
  annotations: {}
  # Service type
  # Ref: https://kubernetes.io/docs/concepts/services-networking/service/#publishing-services-service-types
  type: ClusterIP
  # Ports exposed by the Service.
  # Note: If you want to add additional ports, use .Values.service.extraPorts.
  ports:
    # Port mapped to GraphRAG Chatbot HTTP API.
    http: 8080
  # Exposes the Service on a specific node port on the host machine when "serviceType: NodePort"
  # If left undefined, K8S will pick a random port from the node port range of the cluster.
  nodePort: ""
  # Defines the policy for treating external ingress traffic.
  # By default, Cluster does not preserve client IPs. Change to Local to preserve them.
  # See https://kubernetes.io/docs/tasks/access-application-cluster/create-external-load-balancer/#preserving-the-client-source-ip
  externalTrafficPolicy: ""
  # NodePort used by external load balancers when the external traffic policy is set to Local.
  # By default, Kubernetes will assign a random port, use this to override it.
  healthCheckNodePort: ""
  # Defines the class that should select a particular load balancer implementation.
  # By default, Kubernetes will assign the cluster default implementation, use this to override it.
  # Ref: https://kubernetes.io/docs/concepts/services-networking/service/#load-balancer-class
  loadBalancerClass: ""
  # Source IP ranges for restricting external ingress traffic
  loadBalancerSourceRanges: []
  # External IP addresses at which the Service will be exposed
  # Ref: https://kubernetes.io/docs/concepts/services-networking/service/#external-ips
  externalIPs: []
  # Additional ports to expose with the Service.
  extraPorts: []

##############################
# Persistence Configurations #
##############################

# Configurations for an emptyDir volume where the chatbot configuration files are stored.
# This is used to allow the chatbot to be configured at runtime by NGINX, in a read-only file system.
# To avoid this volume, you need to disable the strict security context, which is not recommended.
# This volume enables an init container that will copy the existing configuration files from the image to the volume.
chatbotConfigVolume:
  # Toggles the config folder emptyDir volume creation.
  # - If enabled, the Deployment will use an emptyDir volume for /usr/share/nginx/html/config/ AND will copy the existing files from the image.
  # - If disabled, the chart won't create and mount ephemeral volumes for /usr/share/nginx/html/config/.
  enabled: true
  # Default location of the config directory inside the container where the empty volume will be mounted.
  mountPath: /usr/share/nginx/html/config/
  # Configurations for an emptyDir volume to be used for /var/cache/nginx.
  # Ref: https://kubernetes.io/docs/concepts/storage/ephemeral-volumes/
  spec:
    # Default emptyDir limit, override to your needs.
    sizeLimit: 1Gi

# Configurations for an emptyDir volume for the Nginx cache folder in each GraphRAG Chatbot container.
# Because the default security context in .Values.securityContext configures the root filesystem to be in read-only mode, certain features
# cannot create and write files in /var/cache/nginx. If you don't use a read-only root filesystem, you can disable this with .Values.nginxCacheTempVolume.enabled
nginxCacheTempVolume:
  # Toggles the temp folder emptyDir volume creation.
  # - If enabled, the Deployment will use an emptyDir volume for /var/cache/nginx.
  # - If disabled, the chart won't create and mount ephemeral volumes for /var/cache/nginx.
  enabled: true
  # Default location of the cache directory inside the container where the empty volume will be mounted.
  mountPath: /var/cache/nginx
  # Configurations for an emptyDir volume to be used for /var/cache/nginx.
  # Ref: https://kubernetes.io/docs/concepts/storage/ephemeral-volumes/
  spec:
    # Default emptyDir limit, override to your needs.
    sizeLimit: 1Gi

# Configurations for an emptyDir volume for the /run folder in each GraphRAG Chatbot container.
# Because the default security context in .Values.securityContext configures the root filesystem to be in read-only mode, certain features
# cannot create and write files in /run. If you don't use a read-only root filesystem, you can disable this with .Values.runTempVolume.enabled
runTempVolume:
  # Toggles the temp folder emptyDir volume creation.
  # - If enabled, the Deployment will use an emptyDir volume for /run.
  # - If disabled, the chart won't create and mount ephemeral volumes for /run.
  enabled: true
  # Default location of the run directory inside the container where the empty volume will be mounted.
  mountPath: /run/
  spec:
    # Default emptyDir limit, override to your needs.
    sizeLimit: 1Gi

#############################
# Deployment Configurations #
#############################

# Configures the strategy of updating Deployment Pods.
# Ref: https://kubernetes.io/docs/concepts/workloads/controllers/deployment/#updating-a-deployment
updateStrategy:
  type: RollingUpdate

# Changes the maximum number of revisions that are kept.
revisionHistoryLimit: 10

# Grace period in seconds before terminating the Pods.
# Ref: https://kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle/#pod-termination
terminationGracePeriodSeconds: 120

# Toggles the auto mounting of API credentials token into the Pods.
# Enable this if you need to contact either the API server or need web identity credentials for federated authentication in cloud APIs.
# Ref: https://kubernetes.io/docs/tasks/configure-pod-container/configure-service-account/#opt-out-of-api-credential-automounting
automountServiceAccountToken: false

# Overrides the default Kubernetes scheduler.
# See https://kubernetes.io/docs/tasks/extend-kubernetes/configure-multiple-schedulers/#specify-schedulers-for-pods
schedulerName: ""

# Overrides the Pod's DNS settings.
# Ref: https://kubernetes.io/docs/concepts/services-networking/dns-pod-service/#pod-dns-config
dnsConfig: {}

# Defines the Pod's policy for DNS resolution.
# Ref: https://kubernetes.io/docs/concepts/services-networking/dns-pod-service/#pod-s-dns-policy
dnsPolicy: ""

# Name of an existing PriorityClass to assign, defining the importance of the pods compared to other pods in the cluster.
# Ref: https://kubernetes.io/docs/concepts/scheduling-eviction/pod-priority-preemption/
priorityClassName: ""

# Overrides the default GraphRAG Chatbot container command.
# Use only for troubleshooting!
# See https://kubernetes.io/docs/tasks/inject-data-application/define-command-argument-container/
command: []

# Overrides the default GraphRAG Chatbot container command's arguments.
# Use only for troubleshooting!
# See https://kubernetes.io/docs/tasks/inject-data-application/define-command-argument-container/
args: []

# Ports used by the GraphRAG Chatbot container
# Note: If you want to add additional ports, use .Values.extraContainerPorts.
containerPorts:
  # Port mapped to GraphRAG Chatbot Nginx listener.
  http: 8080

# Additional labels to append to the Pod definition.
# Values are processed as Helm templates.
# Ref: https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/
podLabels: {}

# Additional annotations to append to the Pod definition.
# Values are processed as Helm templates.
# Ref: https://kubernetes.io/docs/concepts/overview/working-with-objects/annotations/
podAnnotations: {}

###################################
# Security Context Configurations #
###################################

# Defines privilege and access control settings for all containers in the GraphRAG Chatbot Pod.
# See https://kubernetes.io/docs/tasks/configure-pod-container/security-context/#set-the-security-context-for-a-pod
podSecurityContext:
  runAsNonRoot: true
  runAsUser: 10001
  runAsGroup: 10001
  fsGroup: 10001
  fsGroupChangePolicy: OnRootMismatch
  seccompProfile:
    type: RuntimeDefault

# Defines privilege and access control settings for the container running GraphRAG Chatbot.
# See https://kubernetes.io/docs/tasks/configure-pod-container/security-context/#set-the-security-context-for-a-container
securityContext:
  allowPrivilegeEscalation: false
  readOnlyRootFilesystem: true
  capabilities:
    drop: [ "ALL" ]
  seccompProfile:
    type: RuntimeDefault

# Defines privilege and access control settings for the init containers.
# See https://kubernetes.io/docs/tasks/configure-pod-container/security-context/#set-the-security-context-for-a-container
initContainerSecurityContext:
  allowPrivilegeEscalation: false
  readOnlyRootFilesystem: true
  capabilities:
    drop: [ "ALL" ]
  seccompProfile:
    type: RuntimeDefault

#############################
# Scheduling Configurations #
#############################

# Selector labels to match when selecting nodes.
# Values are processed as Helm templates.
# Ref: https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/#nodeselector
# See https://kubernetes.io/docs/tasks/configure-pod-container/assign-pods-nodes/
nodeSelector: {}

# Node and pod affinity & anti affinity configurations for constraining the Pod scheduling.
# Values are processed as Helm templates.
# Ref: https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/#affinity-and-anti-affinity
affinity: {}

# Default podAntiAffinity rule ensuring that GraphRAG Chatbot pods are scheduled on different Kubernetes nodes.
#
# Possible values for .Values.podAntiAffinity.preset are:
# - "soft" (default) - Configures a preferredDuringSchedulingIgnoredDuringExecution rule.
# - "hard" - Configures a requiredDuringSchedulingIgnoredDuringExecution rule.
# Ref: https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/#affinity-and-anti-affinity
podAntiAffinity:
  enabled: true
  preset: soft
  topology: kubernetes.io/hostname

# List of taint tolerations.
# Values are processed as Helm templates.
# Ref: https://kubernetes.io/docs/concepts/scheduling-eviction/taint-and-toleration/
tolerations: []

# Configurations for spreading Pods across different failure domains.
# Values are processed as Helm templates.
# Ref: https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/#pod-topology-spread-constraints
# Ref: https://kubernetes.io/docs/concepts/scheduling-eviction/topology-spread-constraints/
topologySpreadConstraints: []

##########################
# Resource Configuration #
##########################

# Resource configurations for the GraphRAG Chatbot containers.
# Ref: https://kubernetes.io/docs/tasks/configure-pod-container/assign-pods-nodes/
resources:
  limits:
    memory: 1Gi
  requests:
    memory: 1Gi
    cpu: 500m

# Resource configurations for init containers.
initContainerResources:
  limits:
    memory: 16Mi
    cpu: 50m
  requests:
    memory: 16Mi
    cpu: 50m

#########################
# Probes Configurations #
#########################

# Ref: https://kubernetes.io/docs/concepts/configuration/liveness-readiness-startup-probes/

# Configurations for the GraphRAG Chatbot container startup probe.
startupProbe:
  tcpSocket:
    port: http
  failureThreshold: 300
  timeoutSeconds: 1
  periodSeconds: 3

# Configurations for the GraphRAG Chatbot container readiness probe.
readinessProbe:
  tcpSocket:
    port: http
  timeoutSeconds: 5
  periodSeconds: 10

# Configurations for the GraphRAG Chatbot container liveness probe.
livenessProbe:
  tcpSocket:
    port: http
  initialDelaySeconds: 60
  timeoutSeconds: 5
  periodSeconds: 10

########################################
# Additional Deployment Configurations #
########################################

# Additional environment variables to be set for the GraphRAG Chatbot containers.
# Values are processed as Helm templates.
extraEnvFrom: []

# Additional environment variables to be set for the GraphRAG Chatbot containers.
# Values are processed as Helm templates.
extraEnv: []

# Additional volumes to be set for the GraphRAG Chatbot Pod.
# Values are processed as Helm templates.
extraVolumes: []

# Additional volume mounts to be set for the GraphRAG Chatbot containers.
# Values are processed as Helm templates.
extraVolumeMounts: []

# Additional init containers to be inserted after the provisioning init containers.
# Values are processed as Helm templates.
extraInitContainers: []

# Additional GraphRAG Chatbot container ports to expose.
extraContainerPorts: {}

# Additional containers to insert into the GraphRAG Chatbot Pod, e.g. sidecar containers
# Values are processed as Helm templates.
extraContainers: []

########################################
# Pod Disruption Budget Configurations #
########################################

# Configurations for GraphRAG Chatbot default Pod Disruption Budget.
# Ref: https://kubernetes.io/docs/concepts/workloads/pods/disruptions/#pod-disruption-budgets
podDisruptionBudget:
  enabled: true
  minAvailable: 51%
  maxUnavailable: ""

##################################
# Service Account Configurations #
##################################

# Configurations for the default ServiceAccount for GraphRAG Chatbot.
# Ref: https://kubernetes.io/docs/concepts/security/service-accounts/
serviceAccount:
  # Specifies whether a ServiceAccount should be created for GraphRAG Chatbot.
  create: false

  # The name of the ServiceAccount to use.
  #
  # There are three cases to be aware of when using this:
  # - If not set and create is true, a name is generated using the fullname template
  # - If set and create is true, it will use the provided name when creating the ServiceAccount
  # - If set but create is false, it will use the provided ServiceAccount.
  name: ""

  # If .Values.serviceAccount.create is true, insert additional annotations to the created ServiceAccount.
  # Values are processed as Helm templates.
  annotations: {}

#################################
# Additional Kubernetes Objects #
#################################

# Additional objects to insert along with the release.
# Values are processed as Helm templates with tpl function.
# Ref: https://helm.sh/docs/howto/charts_tips_and_tricks/#using-the-tpl-function
extraObjects: []

n8n can be provisioned and updated if needed, by executing SQL scripts against its PostgreSQL database. For this, Graphwise provides the necessary initialization SQL scripts.

Once you have deployed GraphRAG and provided an initialization SQL script, you can use the following command to find the PostgreSQL leader and provision n8n.

CLUSTER_NAME="graphrag-postgres-n8n"
DATABASE_NAME="n8n"
PRIMARY_POD=$(kubectl -n graphrag get pod -l "cnpg.io/cluster=$CLUSTER_NAME,cnpg.io/instanceRole=primary" -o jsonpath='{.items[0].metadata.name}')
kubectl -n graphrag exec -i $PRIMARY_POD -- psql -v ON_ERROR_STOP=1 -d $DATABASE_NAME < n8n_db_script_v.2.4.4.sql

There may be additional steps for completing the n8n workflow integration, such as creating n8n datatables with API keys. This varies by use case, so make sure to follow the documentation for your specific use case.

The GraphRAG Helm chart is designed to be used in a production environment but you need to fine-tune the Helm chart values to match your specific environment and use case. This includes:

To remove your GraphRAG deployment, run the following command:

helm --namespace graphrag uninstall graphrag

There might be leftover resources that need to be cleaned up manually, such as Persistent Volume Claims and their associated Persistent Volumes. This largely depends on your Kubernetes cluster, the persistence settings and related Helm overrides.

You can check for these resources by running the following command:

kubectl -n graphrag get pvc,pv

It is up to you to decide whether to delete these resources or not.