Multi-server-setup upgrade: the master-database' schema didn't got updated
During the upgrade to 3.1p1 I discovered that my master-database schema didn't got updated.
My interface runs on on of my web-servers which also acts as separate server available to the cluster. (So in theory I could put the interface behind a load-balancer, but didn't tested that yet).
- The interface is configured to use the master-database for the interface
- The server is configured to use the master-database as source + a single database for own settings
I didn't researched yet where this happens but I would love to contribute in the following proposal to ensure an idempotent master-db-schema update, independent where I start my multi-server-cluster upgrade. Here is my proposal solution:
- If it's a multi-server setup then check the master-db-version first - if it's lower then perform the master-db schema update
- continue the regular check of the server's-database-version - if it's lower then perform the server-db schema update
- if it's a multi-server setup then also update server.db_version inside the master db (Another Issue #4326 which I will address at the same time)
What do you think?