deploy ha etcd on kubernetes

[raoofm]

I know we have etcd-operator to handle this, but it is not truly functional for prod use cases and doesn't seem to be actively worked on.

It will be great to have a documented way of deploying robust ha etcd cluster on top of kubernetes with/without pv. This should be common usecase that many encounter and can be pretty useful. I was searching on the web but was surprised not to find something documented for such an amazing project.

I think we could leave backup and restore as it could be different for different organizations.

There can be reconfiguration to replace a down node which doesn't respond for configurable time t.

[raoofm]

@gyuho Is there any recommendyed way of deploying etcd on kubernetes? Seems like the operator development has stalled.

[hexfusion]

I am also curious what the future of operator project is. I understand the team is working on other projects currently. I would be interested in getting more involved but the general feeling is this is a less open project and more in-house. People have proposed forking it before, perhaps this would be more realistic now that we are moving to a new org and operator is staying in coreos?

@raoofm maybe we should move this to #9965 ?

[raoofm]

From #9965

We are moving etcd and other sub-projects to its own GitHub organization.

@hexfusion what makes you think operator stays in coreos, i think operator would move along with etcd to the new org.

[hexfusion]

@raoofm per that issue.

Projects that won't be moved:
https://github.com/coreos/etcd-operator
https://github.com/coreos/discovery.etcd.io

[raoofm]

😃 my bad

@hexfusion I agree now

I am also curious what the future of operator project is. I understand the team is working on other projects currently. I would be interested in getting more involved but the general feeling is this is a less open project and more in-house. People have proposed forking it before, perhaps this would be more realistic now that we are moving to a new org and operator is staying in coreos?

[gyuho]

@raoofm @hexfusion "etcd" team is not working on etcd-operator right now. And operator SDK team will be working on etcd-operator.

Seems like the operator development has stalled.

Correct. Nobody is actively working on it. We are looking for ways to revive that project.

[hexfusion]

@raoofm @hexfusion "etcd" team is not working on etcd-operator right now. And operator SDK team will be working on etcd-operator.
[..]
Correct. Nobody is actively working on it. We are looking for ways to revive that project.

@gyuho could you clarify this a little when you get a chance. I guess I don't understand who the "SDK team" is (Red Hat?) . In general, the fact that operator stays with coreos instead of moving to new etcd org makes it feel like a fork would be a reasonable request. This way the etcd org fork can collaborate with SDK team upstream but also continue development and organization on behalf of etcd org community without SDK team interaction.

Please let us know your feeling on a possible fork.

[gyuho]

@raoofm If you can contribute to this, it would be great. No one in etcd team has plan to work on this in the near future.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed after 21 days if no further activity occurs. Thank you for your contributions.