Question regarding Scale Hypercore Configuration

[brucekconvergent]

The typical (smallest) Scale Hypercore cluster consists of 3 nodes -- in the configuration area of VEEAM, where you enter the cluster connection information, you can only enter one host item... is there a supported way to enable VEEAM to use any of the nodes to perform its task (for instance, if the specified node happens to not be available)?

[vnikiforov]

Hello, Bruce,

The wizard takes one address, the DNS name or IP address of any cluster node. You cannot list several nodes.
A second address would not change much with the current design. Per Considerations and Limitations, if any node of the connected cluster becomes inaccessible, backup processing of the whole cluster is interrupted, whichever node you registered. From what I see as per Scale Hypercore documentation any node has full REST API endpoint and there is no dedicated cluster IP so far.

[ronnmartin61]

@brucekconvergent just so you're aware we have had conversations with Scale asking if they would consider implementing some kind of cluster-level virtual IP to address this very issue. As far as we're aware they're not looking to develop this feature however.

[brucekconvergent]

Thanks for the replies guys... means some aggravation if needing to restore something if a node goes missing... I found you can't just add another Scale node, it says it already exists, so I guess the action in that case would be to delete the dead node from veeam config, add the other one in.... I wonder if one can even do that if they have backup jobs defined?

[brucekconvergent]

Hmm... pretty clunky, deleting the scale config deletes the worker, so you have reprovision all of that.
An idea for the VEEAM devs to maybe work around this limitation... maybe let us use a round robin DNS for the node definition on the VEEAM side, and code it so that if it doesn't hit on the first node, it tries the next one, etc. etc. The admin would just need to create a round robin DNS entry in their configuration if this was possible.

[brucekconvergent]

Or... simply update the VEEAM plugin to allow for multiple node entries for failover... probably even simpler.