Thank you Derek for the reply but I disagree with some of the points.
I have Temporal running in K8s cluster as deployment and I have init containers for the pod which I use the same temporalio/server docker image, but run temporal-sql-tool. So it is after the schemas got updated then the new server is deployed. I think there is a little down time right after I upgraded the schema and it’s still the old service running.
I will switch to use temporalio/admin-tools but feels a little strange, and it’s one extra image to download for that pod on that node.