EDIT

The hibernate_sequences are sometimes corrupted (e.g. today it happened in the flora bogota that the AuditEvent that the server tried use was already used in the DB). When I checked the hibernate_sequences table the value was correct but obviously at some time the server did not correctly generate the value either because the original value was not correctly stored in the hibernate_sequences table or not correctly handled in the server.

In the given example finally there were AuditEvent record with id = 21, 22 with a later date (4h) then the records with id 23-30.

We may need both an update to the hibernate_sequences table which exists as script already for the CdmUpdater. And maybe more important an service that forces the server to read the hibernate sequences again and update the internal state.

Maybe this was related to either a problem during creating new user accounts and rights&roles or due to missing rights when trying to add a new taxon, but not sure.

It looks like the hibernate_sequences create problems from time to time so we may also think about not using the advanced identifier strategy anymore. As we have only a single server now it is maybe not required anymore. This needs to be discussed.