Bug with Id: 739013, opened: 4/26/2012
When a table with less than 1000 rows that has an identity value is part of a database that is failed over in an AlwaysOn availability group, the identity is reseeded to 1000. If the identity value is already over 1000 no reseed occurs. This also occurs if you restart the server.
This is happening in Azure databases too.
Example: In a table “X” your Id starts in 1, 2, 3, 4 then suddenly 1005, 1006, etc. And you don’t know when it will happen.
Some people think this has something to do with Sequences like this:
One workaround is using the startup flag T272 but how to specify this in Azure?
Also some people say this keeps happening even when you get over 1000 rows.