Upgrading Multi-Server Configurations

[sproggit]

I have a pair of Technitium DNS Servers set up for a smallish network, with both machines running v14.3.

I've seen the notice for availability of 15.2 and would like to consider an upgrade, but although I've read through the available documentation, I haven't yet found any references to multi-server configurations.

My specific setup provides both authoritative resolution for my local network, plus forwarding services for general internet resolution, which I suspect is pretty common. With a pair of servers, I have one set up as a Primary and one for a Secondary for my local domain (although for reasons I have yet to fathom, my Secondary server keeps dropping out and giving me a "Notify Failed" status message.

Putting that to one side, I'd like to ask if there are any specific actions or sequences of events to be obeyed in a situation where a Primary and a Secondary server are to be upgraded. Is there a "right" and "wrong" sequence? I rather suspect it makes more sense to upgrade the Secondary first, since if that works and a problem with the Primary manifests, I should be able to take the existing primary offline and promote the Secondary to becoming a new Primary...

Unfortunately, the documentation simply doesn't cover this.

Suggestions on a correct upgrade methodology would be gratefully appreciated. Any ideas on how to address "Notify Failed" would also be appreciated.

Thank you!

[ShreyasZare]

Thanks for asking. You can upgrade nodes in the cluster in any order without issues however it is recommended to upgrade the secondary nodes first before upgrading the primary node. This info is included as a note in the install instructions which come as a link in the notification update on the panel.

If primary node is upgraded first and if the data formats have changed then the secondary nodes will generate errors as they wont be able to sync updates from the primary node till they are upgraded too. Other than errors being generated in logs, and the secondary nodes stuck with old data, there is no other side effect.

The notify failed issue you see is not related to clustering. Its just that the secondary nodes received notify request from an IP address it did not recognize as it has another IP set for primary server in its config. You can fix this issue by listing out all IP addresses for all nodes in the cluster and putting them in the "Zone Transfer Allowed Networks" and "Notify Allowed Networks" options in Settings > General section. This will cause all nodes to recognize all IP addresses and make the zone transfer/notify work without any issues.