Loading docs/kubernetes/configure-indexnode.md +40 −0 Original line number Diff line number Diff line Loading @@ -7,6 +7,7 @@ For a full list of available variables, please consult the chart at <!-- TOC depthFrom:2 --> - [Configuring Solr replicas](#configuring-solr-replicas) - [Enabling persistence for Solr instances](#enabling-persistence-for-solr-instances) - [Using external Solr instances](#using-external-solr-instances) <!-- /TOC --> Loading Loading @@ -48,6 +49,45 @@ There are several other variables available in the ESGF Helm chart to customise behaviour - please see the [values.yaml](../../deploy/kubernetes/chart/values.yaml) for a full list of available variables. ## Enabling persistence for Solr instances By default, the ESGF Helm chart configures Solr instances to use local ephemeral storage for the `SOLR_HOME` directories. This means that if a pod gets rescheduled for any reason, all the data stored in that Solr instance is lost, which is (probably) OK for testing but clearly not ideal for a production setup. In order to provide persistent storage for Solr instances, the ESGF Helm chart leverages Kubernetes [Persistent Volumes](https://kubernetes.io/docs/concepts/storage/persistent-volumes/) and [Storage Classes](https://kubernetes.io/docs/concepts/storage/storage-classes/). In this way, the ESGF Helm chart can specify the required storage using [PersistentVolumeClaim](https://kubernetes.io/docs/concepts/storage/persistent-volumes/#persistentvolumeclaims) resources while leaving it up to the cluster operator to attach suitable storage and make it available using a `StorageClass` (the configuration of storage classes is site-specific and beyond the scope of this documentation). To configure persistence for Solr instances, just set the following options: ```yaml solr: persistence: # Enable persistence for Solr instances enabled: true # The storage class to use for Solr volumes # If not given, the default storage class is used storageClassName: fast-ssd # The size of the volume to provision for each type of Solr instance # Defaults to 10Gi for each size: master: 40Gi slave: 40Gi # This is the default size for replica volumes # It can be overridden per-replica if required for large replicas replica: 20Gi ``` There are additional options for advanced storage configurations - please consult the [values.yaml](../../deploy/kubernetes/chart/values.yaml) for a full list. ## Using external Solr instances If you have existing Solr instances that you do not wish to migrate, or need to run Solr Loading Loading
docs/kubernetes/configure-indexnode.md +40 −0 Original line number Diff line number Diff line Loading @@ -7,6 +7,7 @@ For a full list of available variables, please consult the chart at <!-- TOC depthFrom:2 --> - [Configuring Solr replicas](#configuring-solr-replicas) - [Enabling persistence for Solr instances](#enabling-persistence-for-solr-instances) - [Using external Solr instances](#using-external-solr-instances) <!-- /TOC --> Loading Loading @@ -48,6 +49,45 @@ There are several other variables available in the ESGF Helm chart to customise behaviour - please see the [values.yaml](../../deploy/kubernetes/chart/values.yaml) for a full list of available variables. ## Enabling persistence for Solr instances By default, the ESGF Helm chart configures Solr instances to use local ephemeral storage for the `SOLR_HOME` directories. This means that if a pod gets rescheduled for any reason, all the data stored in that Solr instance is lost, which is (probably) OK for testing but clearly not ideal for a production setup. In order to provide persistent storage for Solr instances, the ESGF Helm chart leverages Kubernetes [Persistent Volumes](https://kubernetes.io/docs/concepts/storage/persistent-volumes/) and [Storage Classes](https://kubernetes.io/docs/concepts/storage/storage-classes/). In this way, the ESGF Helm chart can specify the required storage using [PersistentVolumeClaim](https://kubernetes.io/docs/concepts/storage/persistent-volumes/#persistentvolumeclaims) resources while leaving it up to the cluster operator to attach suitable storage and make it available using a `StorageClass` (the configuration of storage classes is site-specific and beyond the scope of this documentation). To configure persistence for Solr instances, just set the following options: ```yaml solr: persistence: # Enable persistence for Solr instances enabled: true # The storage class to use for Solr volumes # If not given, the default storage class is used storageClassName: fast-ssd # The size of the volume to provision for each type of Solr instance # Defaults to 10Gi for each size: master: 40Gi slave: 40Gi # This is the default size for replica volumes # It can be overridden per-replica if required for large replicas replica: 20Gi ``` There are additional options for advanced storage configurations - please consult the [values.yaml](../../deploy/kubernetes/chart/values.yaml) for a full list. ## Using external Solr instances If you have existing Solr instances that you do not wish to migrate, or need to run Solr Loading