Uploaded image for project: 'Confluence Data Center'
  1. Confluence Data Center
  2. CONFSERVER-67891

Request for Azure to have AWS-style Hazelcast cluster join type

    XMLWordPrintable

Details

    • 0
    • We collect Confluence feedback from various sources, and we evaluate what we've collected when planning our product roadmap. To understand how this piece of feedback will be reviewed, see our Implementation of New Features Policy.

    Description

      In situations where it is desired or required to build out your Cloud infrastructure manually (as opposed to using an AWS Quick Start template or the Confluence Data Center Azure Marketplace app to deploy from) you may not initially know what number of peers will be possibly joining or their network addresses. Neither AWS nor Azure support multicast networking, an option which doesn't require a pre-existing list of peers to be defined.

      AWS solves this problem with the "aws" cluster join type option.

       Azure is limited to "tcp_ip" as the only cluster join type available (which requires a list of peers to be defined in confluence.cfg.xml in the confluence.cluster.peers property).

      Request

      Create an "azure" join type that functions like the "aws" cluster join type, is dynamic, and does not require to list all possible peers ahead of time. 

      Workaround

      Create a subnet and DHCP pool dedicated for Confluence and list all possible IP addresses in the Confluence DHCP scope that you have dedicated for Confluence (does not have to be a large pool, 4 Confluence nodes = minimum of 4 IP address in the DHCP scope).

      • this DHCP range/subnet must be routable to the database server and shared attachments directory's host

      When nodes come online, if they are configured with an IP from that range, if their confluence.cluster settings in confluence.cfg.xml define the same cluster name, and if they list each other as peers then they will join up to the same cluster and see each other as members. They will not show the other missing members as being offline, only each other as online and the cluster will not be affected by the absence of the other nodes listed in confluence.cluster settings in confluence.cfg.xml 

      • i.e. if only some and not all listed peers are not online and present, the missing ones will not show as offline 
      • only nodes that had joined recently and then recently went offline would be listed as offline until they drop from the cluster altogether if they do not return

      Attachments

        Issue Links

          Activity

            People

              Unassigned Unassigned
              asmith4@atlassian.com Andrew S
              Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

              Dates

                Created:
                Updated: