#include <db.h> int DB->close(DB *db, u_int32_t flags);
The DB->close() method flushes cached database information to
disk, closes any open cursors, frees allocated resources, and
closes underlying files. When the close operation for a cursor fails,
the method returns a non-zero error value for the first instance of such an error,
and continues to close the rest of the cursors and database handles.
Although closing a database handle will close any open cursors, it is recommended that applications explicitly close all their DBcursor handles before closing the database. The reason why is that when the cursor is explicitly closed, the memory allocated for it is reclaimed; however, this will not happen if you close a database while cursors are still opened.
The same rule, for the same reasons, hold true for DB_TXN handles. Simply make sure you close all your transaction handles before closing your database handle.
Because key/data pairs are cached in memory, applications should make a point to always either close database handles or sync their data to disk (using the DB->sync() method) before exiting, to ensure that any data cached in main memory are reflected in the underlying file system.
When called on a database that is the primary database for a secondary index, the primary database should be closed only after all secondary indices referencing it have been closed.
When multiple threads are using the DB
concurrently, only a single thread may call the DB->close() method.
The DB handle may not be
accessed again after DB->close() is called, regardless of its return.
If you do not close the DB handle explicitly, it will be closed when the environment handle that owns the DB handle is closed.
The DB->close()
method returns a non-zero error value on failure and 0 on success.
The error values that DB->close() method returns include the error values of DBcursor->close() and the following:
A Berkeley DB Concurrent Data Store database environment configured for lock timeouts was unable to grant a lock in the allowed time.
You attempted to open a database handle that is configured for no waiting exclusive locking, but the exclusive lock could not be immediately obtained. See DB->set_lk_exclusive() for more information.
The flags parameter must be set to 0 or be set to the following value:
Do not flush cached information to disk. This flag is a dangerous option. It should be set only if the application is doing logging (with transactions) so that the database is recoverable after a system or application crash, or if the database is always generated from scratch after any system or application crash.
It is important to understand that flushing
cached information to disk only minimizes the window of opportunity
for corrupted data. Although unlikely, it is possible for
database corruption to happen if a system or application crash occurs
while writing data to the database. To ensure that database
corruption never occurs, applications must either: use transactions
and logging with automatic recovery; use logging and
application-specific recovery; or edit a copy of the database, and
once all applications using the database have successfully called
DB->close(), atomically replace the original database with the
updated copy.
Note that this flag only works when the database has been opened using an environment.
The DB->close()
method may fail and return one of the following non-zero errors:
The error messages returned for the first error encountered when DB->close() method closes any open cursors include:
A Berkeley DB Concurrent Data Store database environment configured for lock timeouts was unable to grant a lock in the allowed time.
You attempted to open a database handle that is configured for no waiting exclusive locking, but the exclusive lock could not be immediately obtained. See DB->set_lk_exclusive() for more information.