From eb6bbabb50e4ea4d5bb7449f0ec8815bd3d645fd Mon Sep 17 00:00:00 2001 From: Adi-Beker Date: Sat, 6 Jun 2026 11:50:39 +0300 Subject: [PATCH] docs: clarify same-city dual-site deployment considerations --- docs/en/Deployment.md | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/docs/en/Deployment.md b/docs/en/Deployment.md index 47a1b0587db..d84dcf19f1f 100644 --- a/docs/en/Deployment.md +++ b/docs/en/Deployment.md @@ -2,6 +2,12 @@ ## Cluster Setup +### Same-city dual-site deployment considerations + +For same-city dual-site deployments, avoid treating the two sites as a simple active-active topology without explicit failure-domain planning. NameServer, Broker, and Controller components should be deployed with clear quorum and failover assumptions so that the cluster can tolerate a single-site failure without creating split-brain or unavailable write paths. + +When planning this topology, place replicated Broker groups across independent failure domains, keep NameServer addresses reachable from both sites, and verify that producer and consumer clients are configured with the complete NameServer address list. If automatic failover is required, use a deployment mode that explicitly supports controller-based or replicated high-availability behavior rather than relying only on manual Broker restart or recovery. + ### 1 Single Master mode This is the simplest, but also the riskiest mode, that makes the entire service unavailable once the broker restarts or goes down. Production environments are not recommended, but it can be used for local testing and development. Here are the steps to build.