ukhomeoffice / docker-etcd-backup

Backup process for etcd

Docker Image: ETCD Backup

Cluster Backup process for ETCD v3

Cluster backup

A cluster backup is made with etcdctl backup which strips node information thus requiring a cluster rebuild. This backup is saved as a tar file in the /tmp directory by default. If S3_PATH is specified it will be uploaded to S3 and encrypted with the KMS key that you have specified using the KMS_ID variable.

Configurable options

The following environment variables can be passed in to configure etcd backups for your environment:

ENVIRONMENT VARIABLE	DESCRIPTION	REQUIRED	DEFAULT VALUE
BACKUP_PATH	Set the directory to copy the etcd backup to	N	/tmp
ETCDCTL_CACERT	Verify the ETCD certificate using this CA bundle	N	/srv/kubernetes/ca.crt
ETCDCTL_CERT	TLS certificate file for ETCD	N	/srv/kubernetes/etcd.pem
ETCDCTL_ENDPOINT	ETCD endpoint	N	https://localhost:2379
ETCDCTL_KEY	TLS key file for ETCD	N	/srv/kubernetes/etcd-key.pem
S3_AWS_ENDPOINT	Custom S3 endpoint URL	N	NULL
S3_CA_BUNDLE	The CA bundle for the S3 endpoint	N	NULL
S3_KMS_ID	A KMS ID to use for encrypting the backups in S3	N	NULL
S3_NO_SSL_VERIFY	Skip TLS verify for the S3 endpoint	N	false
S3_PATH	Provide the S3 bucket path to copy the backup to, e.g. s3://my-bucket If unset, the backup is left in BACKUP_PATH	N	NULL

Restore process

1. Stop ETCD everywhere
2. See: https://etcd.io/docs/current/op-guide/recovery/

docker-etcd-backup's Issues

Allow dynamic backup facility

Before a cluster upgrade, we want to be able to take a backup of etcd. Maybe exec into container, touch a file and then etcd backup is taken would be good enough?