Etcd Restore

The Airctl Etcd restore process is essential for restoring the etcd database to a previous state captured in the backup .

Warning

The Airctl Etcd restore feature is added assuming volumes attached at the time of taking backups are in place and intact or preserved by customers by using external support .

Manually copy Etcd backup file on each node at /opt/pf9/pf9-kube/etcd-restore/etcd-backup.db

Command Syntax

The Etcd restore command in Airctl follows a specific syntax:

None

airctl mgmt-etcd-restore --config airctl-config.yaml --verbose

• --config airctl-config.yaml: References the Airctl configuration file, ensuring alignment with the required configurations during restoration.

Detailed Restore Process

Purpose

The restore process aims to replace current Etcd database from a previously captured Etcd snapshot.

Steps Involved

1. Command Invocation

   • Copy etcd backup file manually to each master node at /opt/pf9/pf9-kube/etcd-restore/etcd-backup.db. Required
   • Execute the airctl mgmt-etcd-restore command.
   • Use the --config flag to reference the airctl-config.yaml file, ensuring the restoration aligns with the necessary configurations.
   • Wait for etcd to stablize , check etcd status and member list using below command :

     • /opt/pf9/pf9-kube/bin/etcdctl endpoint status -w table
     • /opt/pf9/pf9-kube/bin/etcdctl member list -w table
   • perform nodelet phases restart on each node by following below steps:

     • systemctl stop pf9-nodeletd.service
     • /opt/pf9/nodelet/nodeletd phases restart
2. Backup Source

   • The Etcd restore command triggers /opt/pf9/pf9-kube/etcd-restore/etcd-restore.sh script shipped via pf9-kube package on each node which uses etcd-backup.db file copied by user on each master node under dir /opt/pf9/pf9-kube/etcd-restore/ .
3. Restoration Process

   • The restoration process uses etcd.env to get ENV variables required to restore etcd cluster .

Usage Example:

Restore Etcd command :

None

airctl mgmt-etcd-restore --config airctl-config.yaml --verbose

Check Etcd status and member list on each node :

None

/opt/pf9/pf9-kube/bin/etcdctl endpoint status   -w table
/opt/pf9/pf9-kube/bin/etcdctl member list   -w table

Perform nodelet phases restart :

None

systemctl stop pf9-nodeletd.service
/opt/pf9/nodelet/nodeletd phases restart

Info

Nodelet phases restart may take 15-20 mins