Skip to Main Content
HCL Domino Ideas Portal

Welcome to the #dominoforever Product Ideas Forum! The place where you can submit product ideas and enhancement request. We encourage you to participate by voting on, commenting on, and creating new ideas. All new ideas will be evaluated by HCL Product Management & Engineering teams, and the next steps will be communicated. While not all submitted ideas will be executed upon, community feedback will play a key role in influencing which ideas are and when they will be implemented.

For more information and upcoming events around #dominoforever, please visit our Destination Domino Page

Status No Plans to Implement
Workspace Domino
Categories Administration
Created by Guest
Created on May 30, 2024

Improve Domino clustering capabilities

In a customer environment, they wanted to explore the possibility of having an active-passive setup using Domino clustering feature, wherein only one node is active in an active-passive HA cluster at any given time. All other nodes act as backups that can readily take over if the active or primary node fails.

The reason behind this is because in the customer's environment, at a particular time, only one node in a cluster is up while the other is down.

So whenever the 2nd node comes up, it will get the delta data from the active node, but the timestamp of the documents in a DB would not be in sync between the two as the "last modified date" will be different on both ends.

Customer is aware that this can be done in OS level clustering, but wanted to have this ability within Domino cluster only.

  • Attach files
  • Admin
    Thomas Hampel
    Reply
    |
    Jun 7, 2024

    Domino clustering is a feature for activly running servers, not for servers that are down.
    Closing this request with no plans to implement

  • Guest
    Reply
    |
    May 30, 2024

    Domino clustering is designed for almost realtime replication not with a stand-by server which needs to be synced perioidically. Domino uses the cluster replicator to push changes immediately when data changes on a server.

    Normal replication should be only configured to periodically replicate in case a cluster push didn't work (if the server was down in that moment, cluster queue did overflow, etc).

    But really Domino clusters are designed to be online and talk to each other replicating almost in real time.
    The streamling cluster replicator is optimized to push data directly from memory instead of reading the data again.

    If you are not real time replicating the other side is not up to date and is not "ready to take over" completely.

    You can set a cluster to restricted to not have users failover due to load balancing. Or set down the availability threshold to only let users access the server when the first server is overloaded or not available.

    About the modified date:
    We need to distinct between the "modified" and "modified in this file" part of a document.
    The modified is the sequence date changed when the document changes. The modified is updated when the document changes.
    Replication uses the modified in this file to check what changed. When replicating data to another server the modified will be the same on both sides. But the modified in this file will differ. That's by design and expected.




    What would make sense is to have better control which primary server to use.
    For mail files the home mail server is preferred by the client. But for applications there isn't a way to have an explicity control where the database is opened. Probably the best you can do today is to set the cluster partner to restricted if only used if needed.

    But this does not solve all the use cases. It would be great to be able to configure it in more detail:

    Define a order of servers to be used (like you want to prefer your server in your branch office etc and only fail over to a remove server).

    Either make it configurable or let the client check based on latency to the server to pick the right cluster mate.
    This way a client would choose the cluster mate that is the closed to the user automatically without any configuration.

    The Notes Client is in general cluster aware and knows about all servers in a cluster. It can also query the Server Availability Index of each server in the cluster with a single request to one server.
    The servers is a cluster keep track about the SAI of all servers in a cluster (check with e.g. "show cluster").

    I don't see that the way you are trying to implement it is what Domino should support.
    But I would see room for improvement in controlling load balancing and choosing the server specially for network optimization with servers in different network locations.


    [ Daniel Nashed / https://blog.nashcom.de ]