The use case is moving a PI DA server to new hardware. In order to minimise data loss and cause minimum disruption to clients, the new server should have the same IP address, hostname/FQDN, and PI System ID (GUID) as the original. It will also be the same version of PI DA - any upgrade, if required, can happen later. To simplify further, I'm just dealing with a single server not a collective.
My question is about interfaces (or any other data source) buffered using PI Buffer Subsystem. It's my understanding that pibufss creates a "session" for each server it's buffering to, and somehow locks all the tags it writes data to (I think the locks are registered by piarchss).
In the ideal scenario, the old server is shut down - forcing pibufss to buffer data - and then the new server comes up looking identical, so pibufss unbuffers all the data. However in the new server, piarchss will not have any of the locks that pibufss thinks it has, so will this cause errors, forcing the user to reset the pibufss session and lose all the buffered data? Or are the locks transient (only while actually writing to a point)?
What if the new server came up with same IP/FQDN but a different PI System ID?