Subject RE: [Firebird-Architect] NBACKUP seriously flawed?
Author Leyne, Sean
Richard,

> It seems to me from my understand of the source code, that when the
> alter database end backup occurs, the write-locking of the database
> header page prevents updates to the rest of the database? (although
not
> reads?)

Actually...

END BACKUP locks the database header for the tiny little moment when it
needs to update backup state there.

Difference file merge operations occur without header lock and allow any
parallel reads/updates to happen (but slowly).

Changes are written both to difference file and to database in this
case, until the difference file merge has been completed.


Sean