--- In ib-support@y..., Helen Borrie <helebor@t...> wrote:

> deployment), many or all users log in as sysdba or owner, then

there is no

> way to know or guarantee that one of those "other sysdbas" who were

already

> logged in when you did the forced shutdown are locked out. They

can just

> keep on chugging away - database integrity be damned.

If other sysdbas ignore the 'maintenance' event post (and what of
users connected but not present to respond to the post?), then
there's not a lot you can do except kill the server as Fred does.

I suppose you could keep checking the connection count and wait until
it is 1 (the backup process itself) before starting the backup...
getting a bit too complicated, though. Ideally this would have been
considered prior to implementing a db where everyone is given sysdba
status. There is no water-tight solution, of course, only best
practices.

Dalton Calford's 'roll-forward logging' seems the best solution so
far for complete failsafe operation:

http://firebird.sourceforge.net/index.php?
op=useful&id=calford_1#rollforward