Shared data between HA nodes

Hello!

The documentation for “Front-end HA Setup” says this about shared data on the nodes:

Shared data between HA nodes
HA deployment requires the filesystem view of most datastores (by default in /var/lib/one/datastores/ ) to be same on all front-ends. It is necessary to setup a shared filesystem over the datastore directories. This document does not cover configuration and deployment of the shared filesystem; it is left completely up to the cloud administrator.

What is the recommended implementation for a shared files system here? NFS share, DRDB, rsync, etc?
Any best practice advice would be highly appreciated!

Thanks,
Thomas

Hi @tku,

You can use the technology more comfortable to you, these datastores need to be shared in order that if one frontend fails the others can have access to the resources needed for keep working (images, …).

It would be enough with a NFS shared between all the nodes.

Hey Christian,

I tried with NFS and it works nicely, thanks.
As a NFS server is a single point of failure I’m going to use a CephFS mount in our existing Ceph Cluster.

Thanks,
Thomas