How replication works in practice?

[ivan.a]

I see that replication has been introduced from the new version which is great news.
I have a couple of questions about this because I've never done this before.

How exactly does replication work? Do secondary nodes work at all times (sync data) or periodically (like backup)?

Will another server immediately take over in case that one server/node failed? Or it will take a time (to boot or something)? So, what exactly happens when a server is down?

 

[wolfgang]

How exactly does replication work?

Periodically (like backup)

Will another server immediately take over in case that one server/node failed?

Replication has noting to do with takeover and does not bring the VM back online.
Replication only duplicate the image to an other node.

So, what exactly happens when a server is down?

Your VM is down and your replication will stop.

see Refmanual

https://pve.proxmox.com/pve-docs/chapter-pvesr.html

 

[ivan.a]

Thank you for reply.

Is there any option for sync data between servers/nodes in v5, so another server/nodes/container can take over in case of failure?
What is best practice?

 

[wolfgang]

There is not storage technology what make a node takeover.
What you mean is HA and HA needs 3 nodes minimum.
Distributed Storage are Ceph, GlusterFS and Sheepdog.
Shared Storage are NFS, iSCSI.