Backup and restore

Velero is a popular open source backup solution for Kubernetes. Its core implementation is a controller running in the cluster that oversees the backup and restore operations. The administrator is given a CLI tool to schedule operations and/or perform on-demand backup and restores. This CLI tool creates Kubernetes resources that the in-cluster Velero controller acts upon. During installation the controller needs to be configured with a repository (called a ‘provider’), where the backup files are stored.

This document describes how to setup Velero with the MinIO provider acting as an S3 compatible object store.

What you will need

  • A running Canonical Kubernetes with DNS enabled

  • MinIO (install described below) or other S3 bucket

  • Velero (install described below)

  • An example workload

Enabling required components

DNS is needed for this setup:

sudo k8s enable dns

Install MinIO

MinIO provides an S3 compatible interface over storage provisioned by Kubernetes. For the purposes of this guide, the local storage component is used to satisfy the persistent volume claims:

sudo k8s enable local-storage

Helm is used to setup MinIO under the velero namespace:

sudo k8s helm repo add minio https://charts.min.io
sudo k8s kubectl create namespace velero

# PRODUCTION MODE
sudo k8s helm install -n velero \
--set buckets[0].name=velero,buckets[0].policy=none,buckets[0].purge=false \
--generate-name minio/minio

Create a demo workload

The workload we will demonstrate the backup with is an NGINX deployment and a corresponding service under the workloads namespace. Create this setup with:

sudo k8s kubectl create namespace workloads
sudo k8s kubectl create deployment nginx -n workloads --image nginx
sudo k8s kubectl expose deployment nginx -n workloads --port 80

Install Velero

Download the Velero binary from the releases page on github and place it in our PATH. In this case we install the v1.14.1 Linux binary for AMD64 under /usr/local/bin:

wget https://github.com/vmware-tanzu/velero/releases/download/v1.14.0/velero-v1.14.0-linux-amd64.tar.gz 
tar -xzf velero-v1.14.0-linux-amd64.tar.gz
chmod +x velero-v1.14.0-linux-amd64/velero
sudo chown root:root velero-v1.14.0-linux-amd64/velero
sudo mv velero-v1.14.0-linux-amd64/velero /usr/local/bin/velero

Before installing Velero into the cluster, we export the kubeconfig file using the config command.

mkdir -p $HOME/.kube
sudo k8s kubectl config view --raw > $HOME/.kube/config

We also export the MinIO credentials so we can provide them to Velero. Be aware that MinIO is used as an S3 bucket replacement for AWS S3, hence the credentials look like they require aws values. This is merely the nomenclature for accessing the S3 bucket in MinIO.

ACCESS_KEY=$(sudo k8s kubectl -n velero get secret -l app=minio -o jsonpath="{.items[0].data.rootUser}" | base64 --decode)
SECRET_KEY=$(sudo k8s kubectl -n velero get secret -l app=minio -o jsonpath="{.items[0].data.rootPassword}" | base64 --decode)
SERVICE=$(sudo k8s kubectl -n velero get svc -l app=minio -o jsonpath="{.items[0].metadata.name}")
cat <<EOF > credentials-velero
[default]
    aws_access_key_id=${ACCESS_KEY}
    aws_secret_access_key=${SECRET_KEY}
EOF

We are now ready to install Velero into the cluster, with an aws plugin that matches the velero release:

SERVICE_URL="http://${SERVICE}.velero.svc:9000"
BUCKET=velero
REGION=minio
velero install \
--provider aws \
--plugins velero/velero-plugin-for-aws:v1.10.0 \
--bucket $BUCKET \
--backup-location-config region=$REGION,s3ForcePathStyle="true",s3Url=$SERVICE_URL \
--snapshot-location-config region=$REGION \
--secret-file ./credentials-velero

Backup workloads

To backup the workloads namespace we use the --include-namespaces argument:

velero backup create workloads-backup --include-namespaces=workloads

Note

Please see the official Velero documentation for details on how to backup persistent volumes and the supported volume types.

To check the progress of a backup operation we use describe, providing the backup name:

velero backup describe workloads-backup 

In the output you should see this operation completed:

Name:         workloads-backup
Namespace:    velero
Labels:       velero.io/storage-location=default
Annotations:  velero.io/resource-timeout=10m0s
              velero.io/source-cluster-k8s-gitversion=v1.30.2
              velero.io/source-cluster-k8s-major-version=1
              velero.io/source-cluster-k8s-minor-version=30

Phase:  Completed


Namespaces:
  Included:  workloads
  Excluded:  <none>

Resources:
  Included:        *
  Excluded:        <none>
  Cluster-scoped:  auto

Label selector:  <none>

Or label selector:  <none>

Storage Location:  default

Velero-Native Snapshot PVs:  auto
Snapshot Move Data:          false
Data Mover:                  velero

TTL:  720h0m0s

CSISnapshotTimeout:    10m0s
ItemOperationTimeout:  4h0m0s

Hooks:  <none>

Backup Format Version:  1.1.0

Started:    2024-07-15 19:35:18 +0000 UTC
Completed:  2024-07-15 19:35:18 +0000 UTC

Expiration:  2024-08-14 19:35:18 +0000 UTC

Total items to be backed up:  18
Items backed up:              18

Backup Volumes:
  Velero-Native Snapshots: <none included>

  CSI Snapshots: <none included>

  Pod Volume Backups: <none included>

HooksAttempted:  0
HooksFailed:     0

Restore workloads

Before restoring the workloads namespace, let’s delete it first:

sudo k8s kubectl delete namespace workloads

We can now create a restore operation specifying the backup we want to use:

velero restore create --from-backup workloads-backup

A restore operation which we can monitor using the describe command is then created:

velero restore describe workloads-backup-20240715193553

The describe output should eventually report a “Completed” phase:

Name:         workloads-backup-20240715193553
Namespace:    velero
Labels:       <none>
Annotations:  <none>

Phase:                       Completed
Total items to be restored:  11
Items restored:              11

Started:    2024-07-15 19:35:54 +0000 UTC
Completed:  2024-07-15 19:35:54 +0000 UTC

Warnings:
  Velero:     <none>
  Cluster:  could not restore, CustomResourceDefinition "ciliumendpoints.cilium.io" already exists. Warning: the in-cluster version is different than the backed-up version
  Namespaces: <none>

Backup:  workloads-backup

Namespaces:
  Included:  all namespaces found in the backup
  Excluded:  <none>

Resources:
  Included:        *
  Excluded:        nodes, events, events.events.k8s.io, backups.velero.io, restores.velero.io, resticrepositories.velero.io, csinodes.storage.k8s.io, volumeattachments.storage.k8s.io, backuprepositories.velero.io
  Cluster-scoped:  auto

Namespace mappings:  <none>

Label selector:  <none>

Or label selector:  <none>

Restore PVs:  auto

CSI Snapshot Restores: <none included>

Existing Resource Policy:   <none>
ItemOperationTimeout:       4h0m0s

Preserve Service NodePorts:  auto

Uploader config:


HooksAttempted:   0
HooksFailed:      0

Listing the resources of the workloads namespaces confirms that the restoration process was successful:

sudo k8s kubectl get all -n workloads

Summing up

Although Velero is a really powerful tool with a large set of configuration options it is also very easy to use. You are required to set up a backup strategy based on the backend that will hold the backups and the scheduling of the backups. The rest is taken care of by the tool itself.