Santhi Kumar Sanapathi:
Hello Everyone. One thing I would like to share here based on my experience from my recent passed CKA exam. I had attempted ETCD backup/restore confidently during middle of my exam itself rather than waiting till the end as others might prefer. Believe me if you understand the concept well, its so easy to answer it. We all know its a sure shot one and a heavy weight one. So, why to skip it? I can strongly say, if you go through both ETCD backup/restore labs in Kodekloud and if you have gone through this - https://github.com/kodekloudhub/community-faq/blob/main/etcd-faq.md and understand it well, then you will be a Rockstar at ETCD question during the exam. The github link mentioned here is a must go through and you should not skip it, especially pay attention to “systemctl list-unit-files | grep etcd”, this has really made my day. Thanks to @Alistair Mackay @unnivkn and Kodekloud team for such a beautiful notes on the ETCD scenarios. I am sorry I cant disclose the exact question I had faced as its against CKA norms. But pls follow what I have suggested. Thanks.
unnivkn:
Hi @Santhi Kumar Sanapathi thanks for your feedback & happy to know that you are able to perform it in your CKA exam.
dexterous neo:
@unnivkn @Santhi Kumar Sanapathi the new ETCD backup/restore lab in Kodekloud which refers to external ETCD server and the one which is at github link with “systemctl list-unit-files | grep etcd”…what is the difference?
Santhi Kumar Sanapathi:
@dexterous neo: “systemctl list-unit-files | grep etcd” is something helpful when you land in a situation where multiple ETCD services/clusters are running from your base node and to identify which is the correct one for your scenario, you need to use this command. If you can go through the GitHub all scenarios as they are very well explained.
Alistair Mackay:
@Santhi Kumar Sanapathi @dexterous neo
Santhi - thanks for your comments 
I have adjusted the terminology slightly on the FAQ page so hopefully it makes a bit more sense.
dexterous neo:
@Santhi Kumar Sanapathi @Alistair Mackay Scenario: “where multiple ETCD services/clusters are running from your base node”. Is there some lab where I can practice this scenario? KodeKloud’s new external ETCD cluster lab has only one External ETCD cluster, in that lab I ssh directly to External ETCD cluster for Backup/Restore. My personal cluster uses stacked ETCD (etcd running as pod). I have no where to try multiple ETCD services/cluster scenario. Please guide.
dexterous neo:
@Santhi Kumar Sanapathi @Alistair Mackay for now I just tried “systemctl list-unit-files | grep etcd” in KodeKloud’s external ETCD lab. I ssh to etcd server and “systemctl list-unit-files | grep etcd” gives me only one etcd.service (instead of multiple, because it has only one etcd, I understood yes), where I note down the IP:Port of --listen-client-urls which matches (obviously) with the one and only --etcd-server’s IP:PORT (which I have already catpured form kube-apiserver.yaml) and then used it in the --endpoint CLIENT_URL where CLIENT_URL is the URL from the --listen-client-urls
Alistair Mackay:
IF there is only one etcd process
AND its port is 2739
THEN you don’t need --endpoints as it will default to the correct one.
Alistair Mackay:
Note you only need certificates and endpoints for snapshot save. Restore simply creates a directory structure from the backup file. You could be having to do a restore because etcd has crashed totally, in which case there would be no service to talk to.
unnivkn:
Hi @dexterous neo multi master etcd is not in the scope of the exam.
ochuko:
@Alistair Mackay can we get a solution video for the <https://kodekloud.com/topic/practice-test-backup-and-restore-methods-2-2/|Practice Test Backup and Restore Methods 2>
unnivkn:
Hi @ochuko please refer the Solution/Hint tab as of now… if you have any questions please let us know. Happy to help.