Active-active Temporal clusters for zero-downtime EKS blue-green migration

Community Support Server Deployment
1

Hi team,
We are planning an EKS blue-green migration (cluster version upgrade) and want to ensure zero downtime for Temporal.

Current setup

  • Temporal is running in the blue cluster

  • Applications are being migrated to the green cluster, currently pointing to Temporal in blue

Goal
During final cutover, we want to:

  • Avoid downtime

  • Continue starting new workflows

  • Preserve all in-flight workflows without disruption

Question
Is it possible to run Temporal in both clusters (blue and green) simultaneously for a short duration (active-active) to enable gradual traffic shifting?

Understanding so far

  • Multi-cluster replication follows an active-passive model

  • Active-active for the same namespace is not supported

  • Sharing a database across clusters may lead to shard ownership and consistency issues

Clarifications

  1. Is there any supported approach to achieve zero-downtime migration in this scenario?

  2. Is running two clusters temporarily against the same database ever safe or recommended?

Thanks!