Distributed Cloud connected version 1.8.0 (current)

For version information, see the Distributed Cloud connected release notes.

Available versions: 1.8.0 1.7.1 1.7.0 1.6.1 1.6.0 1.5.1

Configure Distributed Cloud connected storage

This page describes how to configure Configure Distributed Cloud connected storage, including:

Rakuten's Symcloud Storage
Local storage schemas

Configure Distributed Cloud connected for Symcloud Storage

By default, workloads running on one Google Distributed Cloud connected rack node cannot access the local storage of another Distributed Cloud connected rack node. However, you can configure Distributed Cloud connected racks to use Rakuten Symcloud Storage, which is a third-party solution that acts as a local storage abstraction layer on each Distributed Cloud connected node and makes its local storage available to workloads running on other Distributed Cloud connected nodes. Symcloud Storage is the default and only storage option on Google Distributed Cloud connected servers.

Symcloud Storage is deployed from Google Cloud Marketplace and is subject to the terms stated therein. Google provides limited support for using Symcloud Storage with Distributed Cloud connected and might engage the third-party provider for assistance. Software updates for Symcloud Storage are included in the Distributed Cloud connected software updates.

This release of Distributed Cloud connected ships with and supports Symcloud Storage 5.4.10. No other version of Symcloud Storage is supported in this release of Distributed Cloud connected.

Symcloud Storage classes

This section describe the storage classes that Symcloud Storage can enable on your Distributed Cloud connected cluster. Symcloud Storage on Distributed Cloud connected does not support the robin-rwx storage class nor any custom-configured RWX file system mode volumes. For more information on Symcloud Storage classes, see Using Robin CNS in Kubernetes.

`robin` storage class

The robin storage class is a basic Read Write-Once (RWO) storage class. The following example illustrates the instantiation of the class:

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
    name: robin
    labels:
        app.kubernetes.io/instance: robin
        app.kubernetes.io/managed-by: robin.io
        app.kubernetes.io/name: robin
provisioner: robin
reclaimPolicy: Delete
allowVolumeExpansion: true
volumeBindingMode: WaitForFirstConsumer

`robin-immediate` storage class

The robin-immediate storage class is the same as robin except that the persistent volume is created immediately after creating the corresponding persistent volume claim. The following example illustrates the instantiation of the class:

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
    name: robin-immediate
    labels:
        app.kubernetes.io/instance: robin
        app.kubernetes.io/managed-by: robin.io
        app.kubernetes.io/name: robin
provisioner: robin
reclaimPolicy: Delete
allowVolumeExpansion: true
volumeBindingMode: Immediate

`robin-repl-3` storage class

The robin-repl-3 is an RWO storage class with three replicas that span across multiple Distributed Cloud nodes. The following example illustrates the instantiation of the class:

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
    name: robin-repl-3
    labels:
        app.kubernetes.io/instance: robin
        app.kubernetes.io/managed-by: robin.io
        app.kubernetes.io/name: robin
provisioner: robin
reclaimPolicy: Delete
allowVolumeExpansion: true
volumeBindingMode: WaitForFirstConsumer
parameters:
    replication: "3"
    faultdomain: host

Prerequisites

Before you begin, complete the following steps:

Configure logging and monitoring for the target Distributed Cloud connected project.
Create the target Distributed Cloud connected cluster.
Configure your Distributed Cloud networking so that Pods in the target Distributed Cloud connected cluster can reach the Google Cloud data center.
Bind each local-block persistent volume on each Distributed Cloud node that you don't want to be abstracted by Symcloud Storage. If you unbind a bound local-block persistent volume, then installing Symcloud Storage wipes the contents of that persistent volume. For instructions, see Binding in the Kubernetes documentation.

Install Symcloud Storage on a Distributed Cloud connected node

To install Symcloud Storage on a Distributed Cloud connected node, complete the following steps:

Use the following command to apply the Symcloud Storage license to your cluster. Replace LICENSE_FILE with the full path and name of the Symcloud Storage license file.
```
kubectl apply -f LICENSE_FILE -n robin-admin
```

Use the following command to verify the status of the RobinCluster service and all the Symcloud Storage nodes:

kubectl describe robinclusters -n robinio

The command returns output similar to the following:

[...]
Status:
[...]
Phase:              Ready
robin_node_status:
[...]
 Status:           Ready
[...]
 Status:           Ready
[...]
 Status:           Ready
[...]

The expected status for the service and nodes is Ready.

Set Symcloud Storage as the default storage class

Use the following command to set Symcloud Storage as the default storage class on your Distributed Cloud connected cluster. Replace STORAGE_CLASS with one of the Symcloud Storage classes.

kubectl patch storageclass STORAGE_CLASS -p '{"metadata": {"annotations":{"storageclass.kubernetes.io/is-default-class":"true"}}}'

For more information about setting the default storage class, see Change the default StorageClass in the Kubernetes documentation.

Configure abstracted Symcloud Storage volumes for workloads

This section provides examples for how to use Symcloud Storage classes to configure abstracted storage for your Distributed Cloud connected workloads. For more details about configuring Symcloud Storage volumes, see Using Robin CNS in Kubernetes.

Configure an `ext4` RWO volume in file system mode

The following example illustrates how to configure a persistent volume claim for an RWO volume in file system mode with the ext4 file system. Replace STORAGE_CLASS with one of the Symcloud Storage classes.

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: rwo-fs-pvc
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 10Gi
  storageClassName: STORAGE_CLASS

Configure an RWO volume in block mode

The following example illustrates how to configure a persistent volume claim for an RWO volume in block mode. Replace STORAGE_CLASS with one of the Symcloud Storage classes.

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: rwo-block-pvc
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 10Gi
  storageClassName: STORAGE_CLASS
  volumeMode: Block

Modify the configuration of an existing volume

The following example illustrates how to modify the configuration of an existing Symcloud Storage LZ4 compressed RWO volume by using annotations. ReplaceSTORAGE_CLASS with one of the Symcloud Storage classes.

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: compressed-rwo-fs-pvc
  annotations:
    robin.io/compression: LZ4
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 10Gi
  storageClassName: STORAGE_CLASS

The following example illustrates how to modify the configuration of an existing Symcloud Storage RWO volume with the xfs file system by using annotations. ReplaceSTORAGE_CLASS with one of the Symcloud Storage classes.

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: rwo-xfs-pvc
  annotations:
    robin.io/fstype: xfs
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 10Gi
  storageClassName: STORAGE_CLASS

Configure the Symcloud Storage CLI client

Symcloud Storage provides a command-line interface (CLI) client that you can use to manage your Symcloud Storage configuration. To configure the client on your Distributed Cloud connected cluster, complete the following steps:

Get the Symcloud Storage image path used by the RobinCluster service instance deployed on your Distributed Cloud connected cluster and set your environment variables as follows:

image_robin=$(kubectl get robincluster -o jsonpath='{.items[].spec.image_robin}')
image_registry_path=$(kubectl get robincluster -o jsonpath='{.items[].spec.image_registry_path}')
ROBIN_CNS_IMAGE="$image_registry_path/$image_robin"

Create a robincli resource with the following contents:

kind: Deployment
apiVersion: apps/v1
metadata:
 name: robincli
 namespace: default
 labels:
   name: robincli
spec:
 replicas: 1
 selector:
   matchLabels:
     name: robincli
 template:
   metadata:
     annotations:
       product: robin
     labels:
       name: robincli
   spec:
     containers:
     - name: robincli
       image: ROBIN_CNS_IMAGE
       workingDir: /root
       command: ["/bin/bash","-c","mkdir -p /root/.robin; ln -s -t /usr/lib/python3.7/site-packages/ /opt/robin/current/python3/site-packages/robincli /opt/robin/current/python3/site-packages/stormgr_def.py /opt/robin/current/python3/site-packages/stormgr_lib.py; /opt/robin/current/bin/robin client add-context robin-master.robinio --set-current; while true; do sleep 10000; done"]
       resources:
         requests:
           memory: "10Mi"
           cpu: "100m"

Replace ROBIN_CNS_IMAGE with the full repository path and name of the image that you obtained in step 1.

Apply the robincli resource onto your Distributed Cloud connected cluster.
Upon initial installation, Symcloud Storage generates a default-admin-user secret in the robinio namespace with a randomized password. Use the following commands to obtain these logon credentials:
1. Obtain the user name:
```
kubectl -n robinio get secret default-admin-user -o jsonpath='{.data.username}' | base64 -d
```
2. Obtain the password:
```
 
kubectl -n robinio get secret default-admin-user -o jsonpath='{.data.password}' | base64 -d
```
Log on to the newly created Pod and run the client:
```
kubectl exec -it robincli -- bash
```

Limitations of Symcloud Storage

When using Symcloud Storage with Distributed Cloud connected, you can only achieve high availability if your Distributed Cloud connected cluster consists of three or more Distributed Cloud connected nodes.

Removing nodes that use Symcloud Storage from a cluster

Symcloud Storage volume replicas are stored on worker nodes within your Distributed Cloud connected cluster. If you remove a node from the cluster, the Symcloud Storage volume data stored on that node becomes unavailable. To prevent this, you must do one of the following:

If you are tearing down the entire cluster, remove the workloads and their corresponding Symcloud Storage persistent volumes before you tear down the cluster itself.
If you are removing specific nodes from the cluster, you must migrate the workload data stored on those nodes before you remove those nodes from the cluster. For instructions, see Evacuating volumes from a disk.

Configure local storage schemas

A storage schema is a logical grouping of one or more partitions. Each partition is a logically independent unit of storage. Partitions are created on your cluster sequentially until physical disk space has been exhausted. Each storage schema has a unique name that identifies it.

To create a new local storage schema for your Distributed Cloud connected cluster, you must request it from Google. Once we test the schema and create it on your cluster, you can then apply it using the gcloud CLI.

You cannot modify a schema after it's been applied to a cluster. To change an existing schema, you must request the deletion of the existing schema from Google and then request the creation of a new schema to replace it.

Define partitions for a local storage schema

Before you can request a local storage schema you must first define the partitions for that schema.

A partition has the following properties:

Size. You can either specify a partition size in binary bytes, or have it use all remaining space on the local disk.
Type. You can configure a partition as either a Kubernetes persistent volume (PV) or a Linux local volume on the local disk.
Mode. You can configure the volume stored in the partition as either a block volume or a file system volume. For persistent volume partitions, the partition's storage class is either local-block or local-disks, respectively. For local volume partitions, you can specify the bind and mount points for the contained filesystems.

Request a local storage schema

To request a new local storage schema for your Distributed Cloud connected cluster, contact Google Support and provide the size, type, mode, and, optionally, mount and bind points for each partition you want to create in the schema.

When we receive your request, we run a series of tests to ensure the robustness of the schema, then create it on your Distributed Cloud connected cluster.

Default local storage schemas

Distributed Cloud connected ships with the following default local storage schemas:

default_control_plane_node. This schema defines the following partitions:
- A 100GB local volume partition in file system mode.
- A persistent volume partition in block mode that occupies the remaining free disk space.
default_worker_node. This schema defines a 410GB persistent volume partition in block mode.

Apply a local storage schema to a cluster

To apply a local storage schema to your Distributed Cloud connected cluster, do one of the following:

To apply a local storage schema to the cluster's control plane nodes, use the --control-plane-node-storage-schema flag when creating the cluster. For more information, see Create a cluster.
To apply a local storage schema to the cluster's worker nodes, use the --node-storage-schema when creating a node pool for the cluster. For more information, see Create a node pool.

Distributed Cloud connected creates the partitions defined in your local storage schema upon successful creation of the cluster or node pool.