etcd Backup (Kubeadm)
Assuming the Kubernetes cluster is set up through Kubeadm,
the etcd pods will be running in the kube-system
namespace.
Before taking a backup of the etcd cluster, a Secret needs to
be created in a temporary new or an existing namespace,
containing details about the authentication mechanism used by
etcd. In the case of kubeadm
, it is likely that etcd will
have been deployed using TLS-based authentication. A temporary
namespace and a Secret to access etcd can be created by running
the following command:
$ kubectl create namespace etcd-backup
namespace/etcd-backup created
$ kubectl create secret generic etcd-details \
--from-literal=cacert=/etc/kubernetes/pki/etcd/ca.crt \
--from-literal=cert=/etc/kubernetes/pki/etcd/server.crt \
--from-literal=endpoints=https://127.0.0.1:2379 \
--from-literal=key=/etc/kubernetes/pki/etcd/server.key \
--from-literal=etcdns=kube-system \
--from-literal=labels=component=etcd,tier=control-plane \
--namespace etcd-backup
Note
If the correct path of the server keys and certificate is not provided,
backups will fail. These paths can be discovered from the command that
gets run inside the etcd pod, by describing the pod or looking into the
static pod manifests. The value for the flags etcdns
and labels
should be the namespace where etcd pods are running and etcd pods' labels
respectively.
To avoid any other workloads from etcd-backup
namespace being backed
up, Secret etcd-details
can be labeled to make sure only this Secret
is included in the backup. The below command can be executed to label the
Secret:
$ kubectl label secret -n etcd-backup etcd-details include=true
Backup
To create the Blueprint resource that will be used by K10 to backup etcd, run the below command:
$ kubectl --namespace kasten-io apply -f \
https://raw.githubusercontent.com/kanisterio/kanister/0.96.0/examples/etcd/etcd-in-cluster/k8s/etcd-incluster-blueprint.yaml
Alternatively, use the Blueprints page on K10 Dashboard to create the Blueprint resource.
Once the Blueprint is created, the Secret that was created above needs to be annotated to instruct K10 to use the Blueprint to perform backups on the etcd pod. The following command demonstrates how to annotate the Secret with the name of the Blueprint that was created earlier.
$ kubectl annotate secret -n etcd-backup etcd-details kanister.kasten.io/blueprint='etcd-blueprint'
Once the Secret is annotated, use K10 to backup etcd using the new namespace. If the Secret is labeled, as mentioned in one of the previous steps, while creating the policy just that Secret can be included in the backup by adding resource filters like below:
Note
The backup location of etcd can be found by looking at the Kanister artifact of the created restore point.
Restore
To restore the etcd backup, log in to the host (most likely the Kubernetes
control plane nodes) where the etcd pod is running. Obtain the restore path
by looking into the artifact details of the backup action on the K10 dashboard,
and download the snapshot to a specific location on the etcd pod host machine
(e.g., /tmp/etcd-snapshot.db
). Downloading the snapshot is going to
be dependent on the backup storage target in use. For example, if AWS S3
was used as object storage, the AWS CLI will be needed to obtain the backup.
Once the snapshot is downloaded from the backup target, use the etcdctl
CLI
tool to restore that snapshot to a specific location, for example
/var/lib/etcd-from-backup
on the host. The below command can be used to
restore the etcd backup:
$ ETCDCTL_API=3 etcdctl --endpoints=https://127.0.0.1:2379 \
--cacert=/etc/kubernetes/pki/etcd/ca.crt \
--cert=/etc/kubernetes/pki/etcd/server.crt \
--key=/etc/kubernetes/pki/etcd/server.key \
--data-dir="/var/lib/etcd-from-backup" \
--initial-cluster="master=https://127.0.0.1:2380" \
--name="master" \
--initial-advertise-peer-urls="https://127.0.0.1:2380" \
--initial-cluster-token="etcd-cluster-1" \
snapshot restore /tmp/snapshot-pre-boot.db
All the values that are provided for the above flags can be discovered from the
etcd pod manifest (static pod). The two important flags are
--data-dir
and --initial-cluster-token
. --data-dir
is the
directory where etcd stores its data into and --initial-cluster-token
is the flag that defines the token for new members to join this etcd cluster.
Once the backup is restored into a new directory
(e.g., /var/lib/etcd-from-backup
), the etcd manifest (static pod)
needs to be updated to point its data directory to this new directory
and the --initial-cluster-token=etcd-cluster-1
needs to be
specified in the etcd command argument. Apart from that the volumes
and volumeMounts
fields should also be changed to point to new data-dir
that we restored the backup to.
Multi-Member etcd Cluster
In the cases when the cluster is running a multi-member etcd cluster, the same steps that we followed earlier can be followed to restore the cluster with some minor changes. As mentioned in the official etcd documentation all the members of etcd can be restored from the same snapshot.
Among the leader nodes, choose one that will be used as a restore node and stop the static pods on all other leader nodes. After making sure that the static pods have been stopped on the other leader nodes, the previous step should be followed on those nodes sequentially.
The below command, used to restore the etcd backup, needs to be changed from the previous example before running it on other leader nodes:
$ ETCDCTL_API=3 etcdctl --endpoints=https://127.0.0.1:2379 \
--cacert=/etc/kubernetes/pki/etcd/ca.crt \
--cert=/etc/kubernetes/pki/etcd/server.crt \
--key=/etc/kubernetes/pki/etcd/server.key \
--data-dir="/var/lib/etcd-from-backup" \
--initial-cluster="master=https://127.0.0.1:2380" \
--name="master" \
--initial-advertise-peer-urls="https://127.0.0.1:2380" \
--initial-cluster-token="etcd-cluster-1" \
snapshot restore /tmp/snapshot-pre-boot.db
The name of the host for the flags --initial-cluster
and --name
should be changed based on the host (leader) on which the command is being run.
To explore more about how etcd backup and restore work, this Kubernetes documentation can be followed.
In reaction to the change in the static pod manifest, the kubelet
will
automatically recreate the etcd pod with the cluster state that was backed up
when the etcd backup was performed.