I have problems to take a snaphot backup with: `ETCDCTL_API=3 etcdctl snapshot s . . .

Kubernetes-Slack Discussions
#1

Michael:
I have problems to take a snaphot backup with:
ETCDCTL_API=3 etcdctl snapshot save snapshotdb
after 5 minutes and more I still have an empty snapshot.db.part in the directory. What am I doing wrong? I’ve read the documentation about it, but so far I can’t find what I’m forgetting. I think this is as described in the first part of 128 - ‘Backup and Restore methods’

#2

Mohamed Ayman:
Hello @Michael,
Try the following steps in this link https://github.com/mmumshad/kubernetes-cka-practice-test-solution-etcd-backup-and-restore

#3

Michael:
Ok, thats really. Thank you very much for the help, hopefully I’ll get that in the Kubernetes documentation for the exam.

#4

Michael:
Well, without the paths I could have found this in the k8s documentation with a little more scrolling:
https://kubernetes.io/docs/tasks/administer-cluster/configure-upgrade-etcd/

#5

Akshay Jain:
I hit the same issue once. You’d need to specify the endpoints, cert and key files. Otherwise the command just sits there with an empty .part folder

#6

Michael:
The howto is great, but the last point “After the manifest files are modified ETCD cluster should automatically restart” doesn’t work, the lab ends but the deploys never come back.

#7

Michael:
I have to set this lines: - hostPath:
path: /var/lib/etcd-from-backup --data-dir=/var/lib/etcd-from-backup and - mountPath: /var/lib/etcd-from-backup after that i look with joutnaltctl -xe , but there is no automatic reboot in the lab. I need to run systemctl daemon-reload and
systemctl restart kubelet after that it runs immediatly.

#8

Akshay Jain:
You would need to kill the existing etcd pods so that new ones get created automatically. Alternatively, with you did, the new pods are created when kubelet is alive again.