Hi Alan

> these are the best steps I know to backup/restore cycle.
> 1. shutdown the database (assuming not everyone connected as

SYSDBA - if so

Yes Alan your method is "best" when seen as achieving the safest and
quickest. But shutting down without politely talking to all current
users is not an option. From within each application I will notify
users. The notification will be the response to an event sent from
the FB engine to all connected users because all apps register
interest in an **_AfterInsert event on the table holding a record for
each connected user. No other connections will be allowed as long as
a restore request is connected. When the connected user count has
dropped to 1, meaning the restore requestor is the last conneced
user, that user will disconnect from the datafile without removing
their restore request (thus preventing connections from other than
the user who owns the restore request) and reconnect after the DB
file is restored. Yes the restore would not be to the original
datafile and a delete and rename would occur to prevent having a long
interval wih no datafile.

Regards

Russell

> then you're in trouble - choose a time when noone is connected or

stop the

> server and ignore step 2)
> 2. disconnect as SYSDBA
> 3. rename the database to another name (now noone can connect -

they get

> unavailable database)
> 4. backup the database (start the server again before this if you

had to

> stop it).
> 5. restore to yet another filename.
> 6. rename restored db to original db
> 7. startup db if required.
>
>
> to just backup - do it anytime you like
> Alan
>
>
> <snip>