Kubernetes Setup & Management - Restoring Backups

Backup restoration will only work for Resilient Separated-Planes deployments. If all hosts running etcd fail, follow these steps:

Change your environment type to Cattle. This will tear down the Kubernetes system stack. Pods (the Compute plane) will remain intact and available

Delete reconnecting/disconnected hosts and add new hosts if you need them

Ensure at least one host is labelled etcd=true

For each etcd=true host, mount the network storage containing backups - see Configuring Remote Backups section for details. Then run these commands:

#configure this to point to the desired backup in /var/etcd/backups
target=2016-08-26T16:36:46Z_etcd_1
don’t touch anything below this line
docker volume rm etcd docker volume create --name etcd docker run -d -v etcd:/data --name etcd-restore busybox docker cp /var/etcd/backups/$target etcd-restore:/data/data.current docker rm etcd-restore

Note - you must be logged in as a user with read access to the remote backups. Otherwise, the docker cp command will silently fail

Change your environment type back to Kubernetes. The system stack will launch and your pods will be reconciled. Your backup may reflect a different deployment topology than what currently exists; pods may be deleted/recreated