Skip to content

CHANGELOG.md

Latest commit

 

History

History
688 lines (498 loc) · 29.7 KB

CHANGELOG.md

File metadata and controls

688 lines (498 loc) · 29.7 KB

GraphDB Helm chart release notes

Version 12.3.5

New

Version 12.3.4

New

Version 12.3.3

Fixed

  • Updated all TLS related configurations to use the correct GraphDB v11 properties, see configmap-properties.yaml
  • Updated probes to use the correct HTTPS scheme when TLS is enabled

Version 12.3.2

New

Version 12.3.1

Fixed

Version 12.3.0

New

Version 12.2.1

New

Version 12.2.0

New

Version 12.1.3

New

Version 12.1.2

New

Version 12.1.1

New

Version 12.1.0

New

  • Updated to GraphDB 11.1.0
  • Updated the extraObjects template processing to additionally support rendering string Helm templates.

Version 12.0.3

Fixed

  • Fixed startup failure if no TLS variables are defined, due to misconfiguration of container volumes
  • Replaced raw http usage with $GRAPHDB_PROTOCOL in the indices creation job

Version 12.0.2

New

Version 12.0.1

New

Version 12.0.0

New

  • Updated to GraphDB 11.0.0
  • Added security.overrideUsers that will provision the users on each GraphDB pod startup. Disabled by default.

Updated

  • Tuned the startup and readiness probes to expose the GraphDB service as soon as possible
  • Tuned the startup probe to allow slower GraphDB starts

Version 11.4.1

New

Fixed

  • Fixed the PodDisruptionBudged of the proxy to be deployed only when the proxy is deployed

Version 11.4.0

New

  • Added new configuration properties for GraphDB Tomcat connector SSL/TLS
    • Added configuration.tls.keystore to configure a keystore with its properties
    • Added configuration.tls.truststore to configure a truststore with its properties
    • Added configuration.tls.certificateRevocationList to configure a certificate revocation list
  • Added new configuration properties for configuring GraphDB cluster security (SSL/TLS)
    • Added cluster.tls.mode to configure cluster security mode
    • Added cluster.tls.keystore to configure a keystore with its properties
    • Added cluster.tls.truststore to configure a truststore with its properties
    • Added cluster.tls.certificate to configure a certificate
    • Added cluster.tls.certificateChain to configure a certificate chain
    • Added cluster.tls.certificateKey to configure a private key with its properties
    • Added cluster.tls.rootCerts to configure root certificates to be trusted
    • Added cluster.tls.certificateRevocationList to configure a certificate revocation list
  • Updated jobs and scripts to use https or http depending on whether the Tomcat connector security is configured
  • Added indices configuration enabling a job for initial preloading of indices and other SPARQL updates
  • Added backup.timezone configuration for overriding the default timezone with a specific one

Improvement

  • Added examples for configuring GCP ServiceAccount and Cloud Storage backups under examples/gcp

Version 11.3.5

New

Version 11.3.4

New

Version 11.3.3

Fixed

  • Removed | quote from rendering configuration.properties in properties ConfigMaps in order to allow configuring GraphDB with non-string properties.

Version 11.3.2

New

Version 11.3.1

New

Version 11.3.0

New

  • Added new configuration properties for the license
    • Added license.mountPath to configure where the license volume is mounted
    • Added license.optional to configure the license volume as optional if needed
    • Added license.readOnly to configure the read/write mode of the license volume mount
  • Updated to GraphDB 10.8.0

Updated

  • Removed any pre-install, pre-upgrade, pre-rollback Helm hooks annotations to allow seamless ArgoCD deployments.
  • Changed the license directory to /opt/graphdb/home/conf/license/ with license.mountPath in order to avoid using a subPath volume mount. This allows kubelet to update the license when the Secret has been updated.
  • Changed the license volume mount as read-only by default with license.readOnly

Version 11.2.2

New

Version 11.2.1

New

Version 11.2.0

New

  • Added CronJob for scheduling GraphDB backups. The CronJob supports both local and cloud backups.
    • Added new configurations under backup: backup.enabled for toggling the backup CronJob, backup.type for selecting between local and cloud and more.
    • Local backups support saving the GraphDB backup archives in volume from an existing persistent volume claim, configured with backup.local
    • Cloud backups support uploading the GraphDB backup archives in one of the supported cloud object storage services, configured with backup.cloud
    • Added a new example under examples/backup-local showing how to use the local backup feature with backup.local
  • Added a new property security.provisioner.passwordHash to define an initial password for the provisioner user as a bcrypt hash.
  • Configured graphdb.extra.plugins to load plugins from /opt/graphdb/home/extra-plugins by default

Fixed

  • Updated the GraphDB containers to explicitly use /tmp as a working directory to avoid permission errors due to the default security context's readOnlyRootFilesystem when the container has a starting folder different from /tmp.

Version 11.1.5

New

Improvement

  • Added GraphDB configuration examples
  • Added GraphDB security configuration examples

Version 11.1.4

Fixed

  • Updated StatefulSets templates in graphdb and proxy to properly render extraVolumeClaimtemplates. Replaced if statement with with.

Version 11.1.3

New

Version 11.1.2

New

Improvements

  • Add examples for deploying GraphDB in AWS

Version 11.1.1

New

Version 11.1.0

New

  • Updated to GraphDB 10.7.0
  • Added podAntiAffinity and proxy.podAntiAffinity for configuring a default podAntiAffinity for the GraphDB pods and GraphDB proxy pods. The default values configure a "soft" podAntiAffinity that tries to schedule GraphDB pods across different Kubernetes hosts but does not enforce it.
  • Added new configuration options for the Jobs
    • Added job.schedulerName for overriding the default Kubernetes scheduler
    • Added job.dnsConfig and job.dnsPolicy for customizing the DNS resolution
    • Added job.priorityClassName for defining the pods scheduling importance
    • Added job.nodeSelector, job.affinity, job.tolerations and job.topologySpreadConstraints for customizing the node scheduling
  • Added persistence.volumeClaimRetentionPolicy and proxy.persistence.volumeClaimRetentionPolicy to control the retention policy of the PVCs when the StatefulSets are scaled and deleted. These configurations are used only for Kubernetes 1.27 and above.

Version 11.0.1

GraphDB Helm 11.0.1 is a patch release that includes bug fixes.

Fixed

  • Updated all cluster jobs to explicitly use /tmp as a working directory to avoid permission errors due to the default security context's readOnlyRootFilesystem when the container has a starting folder different from /tmp.
  • Updated all utility scripts to use temporary files under /tmp for the same reason.

Version 11.0.0

Version 11 of the chart addresses a bunch of legacy issues and aims to provide much better user experience and reliability.

Highlights

  • Version - The Helm chart is no longer tied with the version of GraphDB and has a separate development and release cycle.
  • Naming - Removed hardcoded resource names in favor of using the name templates from _labels.tpl
  • Labels - Added the possibility to provide custom labels and annotations to almost every single resource
  • Implementation Agnostic - Removed the dependency of particular ingress controllers and storage classes
  • Security - Enabled security context by default
  • Configurations - Added multiple new configurations to customize both GraphDB and the Kubernetes resources

Breaking

  • Updated the chart to require Kubernetes version 1.26+
  • Enabled security context by default for all pods and containers
  • Updated the GraphDB deployment URL to be http://graphdb.127.0.0.1.nip.io/ by default, see configuration.externalUrl
  • Resource names are no longer hardcoded and are using the templates for nameOverride and fullnameOverride
  • Updated the ingress to be agnostic to the ingress implementation. It will no longer assume that NGINX is the ingress controller in the cluster and will no longer deploy NGINX specific annotations by default. Removed anything related to NGINX as configurations.
  • Removed setting FQDN as hostnames in GraphDB and the proxy in favor of dynamically resolving and configuring the hostnames in the provisioning init containers
  • Removed the default value from global.imageRegistry, the chart now uses the value from image.registry
  • Removed global.storageClass in favor of using by default the default storage class in the cluster. Templates will no longer use global.storageClass.
  • Renamed extraLabels to just labels
  • Moved images.graphdb configurations to just image
  • Moved deployment.imagePullPolicy to image.pullPolicy and deployment.imagePullSecret to image.pullSecrets
    • Note that image.pullSecrets is now a list
  • Moved deployment.ingress to just ingress
  • Moved deployment.tls to ingress.tls
  • Moved graphdb and graphdb.node configurations on the root level
  • Moved all proxy configurations from graphdb.clusterProxy to just proxy
  • Renamed proxy.persistence.enablePersistence toggle to just enabled
  • Moved proxy.serviceType to proxy.service.type
  • Configmaps from graphdb.configs are now under configuration, repositories, cluster and security with a different structure allowing better reuse of existing configmaps
  • Moved graphdb.clusterConfig configurations
    • Moved graphdb.clusterConfig.nodesCount to replicas
    • Moved the rest of graphdb.clusterConfig configurations under cluster, cluster.config and cluster.config.params
  • Moved graphdb.security configurations to security
    • Moved provisioningUsername and provisioningPassword under security.provisioner
  • Moved job related configurations from graphdb (e.g. graphdb.jobResources) to a new root section jobs
  • Moved graphdb.node.service configurations to headlessService
  • Moved graphdb.import_directory_mount configurations to import.volumeMount
  • Renamed pdb to podDisruptionBudget and renamed podDisruptionBudget.create to podDisruptionBudget.enabled for consistency
  • Renamed messageSize to messageSizeKB in the cluster creation configuration in cluster.config.params
  • Renamed java_args to defaultJavaArguments and added a separate javaArguments that can be used for additional configurations, see configuration and proxy.configuration
  • Removed configuration overrides from the default GDB_JAVA_OPTS: enable-context-index, entity-pool-implementation and health.max.query.time.seconds
  • Removed the default logback XML configuration and configmap in favor of an example and a new configuration options under configuration.logback and proxy.configuration.logback
  • Renamed GraphDB storage PVC template name prefix to storage and server import folder to import
  • Moved persistence.volumeClaimTemplateSpec to persistence.volumeClaimTemplate.spec
  • Updated the Service type of the proxy to be ClusterIP by default, see proxy.service.type
  • And more, please refer to values.yaml

New

  • Added GraphDB and GraphDB proxy hostnames resolution in the init containers
  • Added new annotation checksums for GraphDB and GraphDB proxy in order to detect changes in the properties configmaps and ultimately trigger rolling update
  • Added default Secret objects for GraphDB and the proxy that contain sensitive GraphDB configurations
  • Added serviceAccount configurations allowing you to create or use an existing service account for the GraphDB pods
  • Added more feature toggles:
    • headlessService.enabled
    • proxy.service.enabled
    • proxy.headlessService.enabled
    • persistence.enabled
    • proxy.persistence.enabled
    • cluster.jobs.createCluster.enabled - Enables or disables the cluster creation Job
    • cluster.jobs.patchCluster.enabled - Enables or disables the Job for patching the cluster configuration
    • cluster.jobs.scaleCluster.enabled - Enables or disables the Jobs for scaling up or down the cluster
  • Added image.digest to optionally provide an expected digest of the image
  • Added annotations for additional common annotations across all resources
  • Added separate proxy.labels and proxy.annotations configurations for the cluster proxy
  • Added new global.clusterDomain for reconfiguring the default Kubernetes cluster domain suffix in case it is different than cluster.local
  • Added namespaceOverride for overriding the deployment namespace for all resources in case of multi-namespace deployment
  • Added new configuration options for the default ingress ingress:
    • Ability to override the host and path for GraphDB from configuration.externalUrl
    • Ability to change the Ingress path type with ingress.pathType
    • Inserting additional hosts and TLS configurations with ingress.extraHosts and ingress.extraTLS
  • Added security.admin for configuring the initial password of the administrator user
  • Added security.initialUsers.users for inserting additional users into the default initial user.js configuration
  • Added security.provisioner.existingSecret and security.provisioner.tokenKey to provide an existing authentication token
  • Added cluster.token.existingSecret and cluster.token.secretKey for using an existing Secret instead of providing the cluster secret token as plaintext in values.yaml
  • Added cluster.config.existingConfigmap to specify a custom configmap key if needed
  • Added configuration.properties and proxy.configuration.properties for appending additional inline GraphDB configurations in their properties configmaps
  • Added configuration.secretProperties and proxy.secretProperties for appending additional inline sensitive GraphDB configurations if needed
  • Added configuration.extraProperties.existingConfigmap and proxy.configuration.extraProperties.existingConfigmap for appending GraphDB properties from an existing ConfigMap resource
  • Added configuration.extraProperties.existingSecret and proxy.configuration.extraProperties.existingSecret for appending GraphDB properties from an existing Secret resource
  • Added a Service for single GraphDB deployments, configured with new configurations under service
  • Added new configurations for the Service resources service, headlessService, proxy.service and proxy.headlessService:
    • Added labels configurations for insertion of additional labels
    • Added ports mappings in each Service
    • Added extraPorts for mapping additional ports, use in combination with extraContainerPorts
  • Added containerPorts and proxy.containerPorts for mapping the ports on which GraphDB listens on
  • Added extraContainerPorts and proxy.extraContainerPorts to open additional container ports
  • Added service.externalTrafficPolicy and service.proxy.externalTrafficPolicy to override the policy to Local if needed
  • Added service.healthCheckNodePort and service.proxy.healthCheckNodePort to define a specific node port for LB health checks
  • Added service.loadBalancerClass and service.proxy.loadBalancerClass to select a specific load balancer implementation
  • Added service.loadBalancerSourceRanges and service.proxy.loadBalancerSourceRanges to restrict the external ingress traffic from the LB
  • Added service.externalIPs and service.proxy.externalIPs to use existing external IPs
  • Added persistence.emptyDir and proxy.persistence.emptyDir configurations for an emptyDir volume that will be used when the persistence is disabled
  • Added tempVolume configurations for an emptyDir volume mapped to the /tmp folder in the GraphDB containers
  • Added configurations for extra labels and annotations for all persistent volume claim templates: persistence.volumeClaimTemplate, proxy.persistence.volumeClaimTemplate and import.volumeMount.volumeClaimTemplate
  • Added imagePullPolicy configuration to the Jobs containers
  • Added jobs.backoffLimit for configuring the retry count for all jobs
  • Added jobs.ttlSecondsAfterFinished for configuring the time in seconds for all jobs before deleting finished pods
  • Added jobs.persistence.emptyDir configurations for the default temporary storage for all jobs
  • Added proxy.command and proxy.args that override the default container entrypoint and command, use for troubleshooting
  • Added proxy.pdb for configuring a pod disruption budget for the GraphDB Proxy
  • Added proxy.logback configurations for providing the proxy with a custom Logback XML configuration
  • Added proxy.initContainerSecurityContext and proxy.initContainerResources to avoid using the configurations from GraphDB
  • Added automountServiceAccountToken with default value false effectively ejecting the service account token by default
  • Added updateStrategy and proxy.updateStrategy for controlling the strategy when updating pods
  • Added podManagementPolicy and proxy.podManagementPolicy for configuring how the pods are created and scaled
  • Added schedulerName and proxy.schedulerName for overriding the default Kubernetes scheduler
  • Added dnsConfig, dnsPolicy, proxy.dnsConfig and proxy.dnsPolicy for customizing the DNS resolution if needed
  • Added extraContainers and proxy.extraContainers for inserting additional containers into the pods of GraphDB and the GraphDB proxy
  • Added initContainerDataPermissions and proxy.initContainerDataPermissions for changing permissions in the storage volumes if needed
  • Added extraVolumeClaimTemplates and proxy.extraVolumeClaimTemplates
  • Added extraObjects as a way to insert additional Kubernetes objects into the deployment
  • Added priorityClassName and proxy.priorityClassName configurations

Updates

  • GraphDB and GraphDB proxy properties configmaps are now applied by default
  • References to existing configmaps and secrets are now processed as templates
  • Node scheduling configurations are now processed as templates
  • Values in labels, annotations and imagePullSecrets are now evaluated as templates
  • Removed unused busybox image configurations from images.busybox
  • Renamed the port mappings of GraphDB and GraphDB proxy to http and rpc
  • Service resources and probes now refer to the target ports by their nicknames instead of explicit port numbers
  • Added trimming when loading files in the configmaps and secrets
  • Cluster jobs now automatically resolve the cluster domain
  • Removed files/config/graphdb.properties and files/config/proxy/graphdb.properties and moved any defined properties directly into the ConfigMap declarations
  • Moved GraphDB specific properties from GDB_JAVA_OPTS into the properties ConfigMaps
  • Added -XX:-UseCompressedOops in the default Java arguments to allow allocating heap sizes larger than 32GBs when the max heap size is based on the -XX:MaxRAMPercentage Java option
  • Ejected the default service account token in the GraphDB proxy pods
  • Overhauled NOTES.txt to be more helpful
  • Added default resource limits and requests for all init containers and provisioning jobs
  • PodDisruptionBudget are enabled by default for both GraphDB and GraphDB proxy
  • Updated init containers to invoke bash instead of sh
  • Updated the default memory limits and requests to 4Gi

Version 10.6.0-R2

New

  • Added graphdb.node.extraInitContainers and graphdb.clusterProxy.extraInitContainers that allows for the insertion of custom init containers to both GraphDB and its proxy
  • Added graphdb.clusterConfig.transactionLogMaximumSizeGB configuration for the cluster creation JSON configuration.
  • Added graphdb.clusterConfig.existingClusterConfig for providing a custom cluster creation JSON configuration.

Fixed

  • Fixed URLs in the README.md that refer to the official GraphDB documentation.
  • Fixed the cluster creation JSON configuration to use messageSizeKB instead of messageSize, see graphdb.clusterConfig.messageSize.

Version 10.6.0

New

  • Added graphdb.node.licenseFilename for cases where the default filename is not "graphdb.license"

Improvements

  • Updated the default ingress's path type to ImplementationSpecific
  • Updated graphdb.properties example file
  • Templates will now use Chart.AppVersion by default unless images.graphdb.tag is specified.
  • Updated busybox image to version 1.36.1
  • Use clusterCreationTimeout in patch cluster job as well

Version 10.5.1-R2

  • Fixed graphdb-cluster-proxy-configmap to use the correct java_args configuration from values.yaml.

Version 10.4.1

  • Added configurations for specifying resource values for all remaining containers, see graphdb.node.initContainerResources and graphdb.jobResources.

Version 10.3.1-R2

Improvements

  • Fixed the image registry to have priority over the global registry

Version 10.3.0-R2

New

  • Added configurations for extra service annotations, see graphdb.node.service.annotations, graphdb.clusterProxy.service.annotations and graphdb.clusterProxy.headlessService.annotations

Version 10.2.3

New

  • Added configurations for overriding graphdb-node's command and arguments, see graphdb.node.command and graphdb.node.args
  • Added configurations for Pod Disruption Budget for the GraphDB nodes, see graphdb.pdb
  • Added graphdb-proxy-properties-configmap.yaml to load graphdb.properties containing the cluster node addresses into the cluster-proxy

Changed

  • Removed versions field as it is not really used nor needed
  • Removed the license provisioning init container in favor of directly mounting the license
  • Removed unused graphdb-node-storage volume mount
  • Removed the node addresses from the graphdb-cluster-proxy-configmap.yaml to prevent cluster proxy restarting on cluster scale up/down
  • Updated the resources to not set CPU limits in order to avoid CPU throttling, lowered the default CPU requirements

Version 10.2.2

New

  • Added configurations for extra env vars in the nodes and cluster proxies, see graphdb.node.envFrom and graphdb.clusterProxy.extraEnv.
  • Added configurations for changing the revisionHistoryLimit for nodes and cluster proxies.
  • Added configurations for adding extra podLabels and podAnnotations for both the nodes and cluster proxies.
  • Added configurations for terminationGracePeriodSeconds to both the nodes and cluster proxies.
  • Fixed an issue with setting the provisioningUsername to anything other than the default.

Improvements

  • Updated the templates to avoid rendering empty configurations
  • Removed unused helper template graphdbLicenseSecret
  • Added graphdb prefix in the helper templates function naming

Version 10.2.1

New

  • Added configurable security context for both the node and cluster-proxy statefulsets and all the jobs
  • Added extraEnv, extraVolumes and extraVolumeMounts to the statefulsets
  • Added an optional PV/PVC to the cluster-proxy to properly preserve logs (enabled by default)
  • Changed the provision user credentials to be used through a secret instead of rendering inside the jobs
  • Changed the logback.xml and graphdb.properties provisioning to work even if such are already present
  • Changed the graphdb-cluster-config-configmap map to not render when there is no cluster
  • Changed the default values of nodeSelector, affinity, tolerations and topologySpreadConstraints to be a part of the values.yaml file instead of inside the statefulsets
  • Updated default clusterConfig.electionMinTimeout and clusterConfig.electionRangeTimeout to the current GraphDB defaults
  • Updated the cluster proxy probes settings, so it can become available sooner
  • Updated the cluster and repositories jobs with simpler arguments removing the need to copy scripts and to make them executable
  • Added ephemeral volumes in the cluster and repositories jobs to avoid issues with readonly file systems

Version 10.2.0-R2

New

  • Added the ability to provision a repository

Version 10.1.5-R2

New

  • Fixed an issue with the external proxy connecting to the nodes when https is used

Version 10.1.2-R2

New

  • Added ability to override cluster proxy's type, default remains LoadBalancer

Version 10.1.1-R2

New

  • Fixed ingress template to properly handle root context
  • Fixed single node returning wrong location header with explicit transactions

Version 10.0.1

Breaking

  • The graphdb-node service now is always headless. If you installed Version 10.0.0 with graphdb.clusterConfig.nodesCount set to 1 you will have to delete the service prior to an update

New

  • Upgrade to GraphDB 10.0.1
  • Cluster size can now be scaled
  • Fixed an issue with deploying with security turned on
  • Fixed an issue with the cluster proxy returning its internal address when queried externally

Version 10.0.0

Breaking

New major release that isn't compatible with the old chart, due to major breaking changes in Graphdb 10. Migration steps can be found here.

New

  • Changed to work with the new GraphDB 10.
  • Removed Kong.
  • Moved from multiple stateful sets with 1 replica to statefulsets with multiple replicas.
  • Configurable liveness, readiness, startup probes.
  • Can use standalone without license by default. Don't forget to set your license for a working cluster and connectors!
  • New overridable configmaps for users, settings and logback.

Version 9.9.0

Breaking

  • images.graphdb, images.kong and images.busybox are now maps which can specify registry, repository and tag

New

  • Added global variables support (global.deployment.host/global.ingressHost, global.storageClass, global.imagePullSecrets and global.imageRegistry)
  • Add ability to override logback.xml by setting deplyment.logbackConfigFile to the location of the file to use
  • Set additional JMX attributes using graphdb.masters.additionalJmxArrtibutes. This is a map of attr_name=attr_value pairs
  • Fixed loadrdf tool path
  • Moved to dynamic volume provisioning by default (volumeClaimTemplates), old default pvc/pv's are still available
  • Added JDBC driver support for Ontop functionality
  • Minor fixes

Version 9.8.1

New

  • Added multiple repositories provisioning
  • Added security provisioning
  • Added GraphDB properties provisioning
  • Changed GraphDB vhosts and external url properties
  • Upgrade to GraphDB 9.8.1
  • Provide flexible persistence provisioning
  • Provide HA options like node selectors, podaffinity, tolerations, etc
  • Make Ingress and kong optional
  • Minor fixes