Hunter Lee

@hanm Thanks for the follow-up! This is helpful. Just so I don't leave you in the dark, I am actively reviewing this. There are a few things that I think...

One thing I forgot to mention is to add some sort of leader-election protocol for redundancy among the nodes whose `backup.enabled` is set to `true`. I know there's a comment...

@hanm thanks for the reply. I'll keep the discussion on this PR then. I realized from the email that the JIRA tickets were getting updated automatically as well. > Adding...

Thanks @hanm . Also a question here if you could clarify - some parts of the code are missing so I'm just making educated guesses. For `SnapBackup`, is my understanding...

Update: We have a variant of the backup and recovery support initially shared by @hanm . We've been running it at LinkedIn for some time with success. Some code has...

It is currently publicly available here in LinkedIn's public fork of ZooKeeper: https://github.com/linkedin/zookeeper/commits/li-dev/release-3.6.2-1-backup We run an additional observer node dedicated for backup purposes and use an internal NFS-based cold storage...

Published a PR at https://github.com/apache/zookeeper/pull/1883. I just need to do another pass but I think this commit includes all changes. Let me know if there's anything missing or off anywhere.

@li4wang Great. I think you can also do an on-demand backup/restore with just slight modifications on https://github.com/apache/zookeeper/pull/1883.