Questions about upcoming clustering support

[stratself]

Thank you Shreya for writing a sneak peak post on Reddit for the clustering feature :)

With a release coming soon, I'm thinking of reinstalling Technitium from scratch in my homelab. I hope to ask about a few things in preparation for this task (I'm not on Reddit):

1. What's the protocol being used to communicate between nodes in the cluster? I believe you hinted at HTTP in the past, and if so, should I utilize TLS certs to encrypt inter-cluster communication?

2. Could the nodes be configured to sync only some zones, rather than all zones? This is needed as I'd like to set per-node FWD records for the default root zone at . (somewhat similar to #1382 comment), while also syncing certain intranet-only domains and their DNSSEC failover within the cluster.

3. Lastly, regarding the "single admin panel access will allow you to log into any node" bit. Can node-specific data such as those described in 2. be edited with this admin panel access? Furthermore, could it be configured that only some nodes can write data, as a form of access control?

I hope these questions are okay at this stage, though feel free to ignore them should you want to annnounce everything later. Once again, thank you so much for your work in clustering as well as TDNS as a whole!

[ShreyasZare]

Thanks for asking. The clustering uses HTTP API that the DNS server already has for all other things. The only difference is that the API calls between two servers are HTTPS only so either you have to configure TLS for the web service manually before initializing clustering or it will enable it automatically with self signed cert by default.

The zones function just the same as they do not. The clustering feature just adds a Catalog zone which gets synced across all nodes and only zones that you add to the cluster catalog zone will get replicated across all cluster nodes. Rest all things remain just like they are and you can create your own zones or Catalog zones as required.

The single panel admin access will be such that you will have a drop down option to select between nodes and it will automatically show you data from that specific node. So you get to access all nodes by logging into any of the admin panels. The config can only be updated on primary node and gets synced to secondary nodes, which is similar in architecture to how zones replicate. Only a few options which are specific to the node, like local end points to listen on or TLS certs etc. will be separately handled but will be available on the same admin panel you login to.

Let me know if you have any more queries.

[stratself]

Thanks for the detailed response. I didn't know v13 and v14 are closely related to each other like that - this makes it easier to understand the previous blogpost about Catalog Zones.

I've forgot to touched on an important aspect: blocklists/allowlists. Are they also synced, and how? From reading previous posts, I have a hunch that they're also implemented as zones, so would they be easily included inside the cluster's catalog zone?

[ShreyasZare]

The Catalog zones feature was always planned to be used with Clustering. It was implemented earlier so that any issues with it will be worked out before Clustering is implemented.

Yes, everything including allow/block lists, Apps, Settings, etc. will be synced across all nodes. This is done using a modified backup and restore implementation where the primary node creates a zip file based on things that are updated and it is restored by secondary nodes. This ensures that only changes that were made from the last sync time are transferred to avoid sending large amount of data each time for every small change. But the design is essentially taking a backup on the primary and restoring it on secondary. Only zones part is left out of this mechanism and it relies totally on the Catalog zone.

[stratself]

Hi again, I really appreciate your answers

I'd also wanna ask that: will the promotion of Secondary node to Primary node (in case of failures) be automatic or does it require some manual intervention? Also, is it feasible to support multiple Primary nodes in the future?

[ShreyasZare]

The promote to primary is not automatic at the moment but option will be added in later updates to define rules for doing this. Supporting multiple primary nodes is not feasible due to how the DNS zones work.