Yeah, I think I've got it sorted at this point. Keepalived is configured on both, master registers as master and backup picks up when master goes offline.

I am having one issue though - it seems any time keepalived does anything on the backup server, it shuts down all the containers that are running.

27 minutes ago, VanGoghComplex said:

I am having one issue though - it seems any time keepalived does anything on the backup server, it shuts down all the containers that are running.

The containers on the Backup server will only run when the Main server is online.

I'm sorry, I wasn't clear - non-replicated containers on the backup server are shutting down whenever the master goes offline.

My backup server is a game server, not a dedicated backup, so it's running some other stuff - Teamspeak, Crafty and the like. I do see the replicated containers in the list where they belong. But when the master goes offline, the backup seems to shut down all running containers.

57 minutes ago, VanGoghComplex said:

I'm sorry, I wasn't clear - non-replicated containers on the backup server are shutting down whenever the master goes offline.

How does the configuration look like, what does the syslog say, did it possibly say that the MASTER server isn't even online? Maybe this is a keepalived issue.

I'm not too sure, the plugin is used by a few folks and it seems to work everywhere.

1 hour ago, VanGoghComplex said:

My backup server is a game server, not a dedicated backup, so it's running some other stuff - Teamspeak, Crafty and the like. I do see the replicated containers in the list where they belong. But when the master goes offline, the backup seems to shut down all running containers.

Oh, this could be the issue, the plugin is specifically designed to be "replication" like.

1 hour ago, ich777 said:

How does the configuration look like, what does the syslog say, did it possibly say that the MASTER server isn't even online? Maybe this is a keepalived issue.

I'm not too sure, the plugin is used by a few folks and it seems to work everywhere.

Oh, this could be the issue, the plugin is specifically designed to be "replication" like.

Gotcha. That makes a little sense.

It seems like any time the replicate finishes, or when the master goes offline, the backup shuts down all running containers.

Is it possible that the plugin might get an update for this in the future, or is it a technical challenge?

12 hours ago, VanGoghComplex said:

It seems like any time the replicate finishes, or when the master goes offline, the backup shuts down all running containers.

When the MASTER goes offline all containers are shutting down? Shouldn't it be the complete opposite?

12 hours ago, VanGoghComplex said:

Is it possible that the plugin might get an update for this in the future, or is it a technical challenge?

To only start stop containers which are replicated? That would be indeed be possible but not planed since this can cause some other issues and not the intended use case for this plugin.

That's a bummer. So what's happening is:

Master Server:
Non-replicated containers: UP
Replicated containers: UP

Backup Server:
Non-replicated containers: UP
Replicated containers: DOWN

!!! EVENT - MASTER OFFLINE !!!

Master Server: OFFLINE

Backup Server:

Non-replicated containers: UP
Replicated containers: UP

!!! EVENT - MASTER ONLINE !!!

Master Server:
Non-replicated containers: UP
Replicated containers: UP

Backup Server:
Non-replicated containers: DOWN
Replicated containers: DOWN

The plugin is stopping all containers on the backup server once the master becomes available.

This is a minor inconvenience for me because my backup server is a gaming server and isn't running anything important. However, it is an annoying additional touchpoint after a failover. The stopped non-replicated containers do not come back up even if set for --restart always.

Maybe my use case is unique, but the idea of having two Unraid servers in your setup and hosting nothing but backups on the second feels odd.

25 minutes ago, VanGoghComplex said:

The plugin is stopping all containers on the backup server once the master becomes available.

Yes, that was a safety implementation to make sure all containers get shutdown, I'm not sure if that would be easily implementable since the Replication plugin was designed to replicate the MASTER server so if it goes down the SLAVE server could spin up containers for a short period of time to bridge times of short downtime from the MASTER server.

28 minutes ago, VanGoghComplex said:

The stopped non-replicated containers do not come back up even if set for --restart always.

Yes, because the plugin stops the containers and therefore --restart will not work because basically they where stopped intentionally, at least from the plugins perspective.

29 minutes ago, VanGoghComplex said:

Maybe my use case is unique, but the idea of having two Unraid servers in your setup and hosting nothing but backups on the second feels odd.

I don't think so, but the use case for the plugin is as I've described, to bridge short downtime from the MASTER server.

However you could write a small script that starts the game server containers when the MASTER server is going online again, just look in your keepalived.conf file on the SLAVE server.

Not sure why the vm backup is disable and grayed out. i got containers replicating just fine. Any ideas on what to check? Source & Destination are ZFS and same configuration for default vm disk location /mnt/nvme/virtual_machines on both source and dest.

3 hours ago, UnraidUser_2020 said:

Not sure why the vm backup is disable and grayed out.

VM ? as only Dockers and / or LXC Containers are supported.

3 hours ago, UnraidUser_2020 said:

Any ideas on what to check?

may take a look at the logs when running the task, you can start in unraid terminal by executing

unraid-replication