I’m wondering how Tamino handles the backup procedure internally, because to take a backup procedure the database should be active. This means that during the backup procedure users can make changes to the databases while Tamino is making a backup of it.
I’ve been executing tests and it seems that during the backup procedure no locks are set to prevent users making updates during backup. Is that so? Is there some technical doc describing these kind of issues?
As I understand it, all modifications to Tamino data take place in memory and are periodically written physically back to Tamino with a buffer flush. As far as a backup goes it initially backs up the physical database whilst updates are still taking place in memory. Then it triggers a buffer flush which writes changes back to Tamino and to the backup. A new log file is also started. The backup contains a full save of the database (including the changes that took place while it was running). This means the database can be restored right up to the point the new log was started.
I’ve seen that Tamino starts a database from time to time in standby mode (for example dduring restore). In that case users can’t connect to the database until the database is restarted in normal mode. Can I start a database in standby mode via argbatch?
Standby mode is when Tamino is doing internal processing, for example when it is applying logs, rolling forward from a backup. It’s not possible for a user to start a database in standby mode.