#include <db_cxx.h> int DbEnv::txn_begin(DbTxn *parent, DbTxn **tid, u_int32_t flags);
The DbEnv::txn_begin() method creates a new transaction in the
environment and copies a pointer to a DbTxn that
uniquely identifies it into the memory to which tid
refers. Calling the
DbTxn::abort()
,
DbTxn::commit()
or
DbTxn::discard()
methods will
discard the returned handle.
Transactions may only span threads if they do so serially; that is, each transaction must be active in only a single thread of control at a time. This restriction holds for parents of nested transactions as well; no two children may be concurrently active in more than one thread of control at any one time.
Cursors may not span transactions; that is, each cursor must be opened and closed within a single transaction.
A parent transaction may not issue any Berkeley DB operations — except for
DbEnv::txn_begin(),
DbTxn::abort()
and
DbTxn::commit()
— while it
has active child transactions (child transactions that have not yet
been committed or aborted).
The DbEnv::txn_begin()
method either returns a non-zero error value or throws an
exception that encapsulates a non-zero error value on
failure, and returns 0 on success.
If the parent parameter is non-NULL, the new transaction will be a nested transaction, with the transaction indicated by parent as its parent. Transactions may be nested to any level. In the presence of distributed transactions and two-phase commit, only the parental transaction, that is a transaction without a parent specified, should be passed as an parameter to DbTxn::prepare() .
The flags parameter must be set to 0 or by bitwise inclusively OR'ing together one or more of the following values:
DB_READ_COMMITTED
This transaction will have degree 2 isolation. This provides for cursor stability but not repeatable reads. Data items which have been previously read by this transaction may be deleted or modified by other transactions before this transaction completes.
DB_READ_UNCOMMITTED
This transaction will have degree 1 isolation. Read operations
performed by the transaction may read modified but not yet committed
data. Silently ignored if the DB_READ_UNCOMMITTED
flag was not specified when the underlying database was opened.
Enable transactional bulk insert optimization. When this flag is set, the transaction avoids logging the contents of insertions on newly allocated database pages. In a transaction that inserts a large number of new records, the I/O savings of choosing this option can be significant.
Users of
this option should be aware of several issues. When
the optimization is in effect, page allocations that
extend the database file are logged as usual; this
allows transaction aborts to work correctly, both
online and during recovery. At commit time, the
database's pages are flushed to disk, eliminating the
need to roll-forward the transaction during normal
recovery. However, there are other recovery
operations that depend on roll-forward, and care must
be taken when DB_TXN_BULK transactions interact with
them.
In particular, DB_TXN_BULK is incompatible
with replication, and is simply ignored when
replication is enabled. Also, hot backup procedures
must follow a particular protocol, introduced in
Berkeley DB 11gR2.5.1, which is to set the
DB_HOTBACKUP_IN_PROGRESS
flag in the environment before
starting to copy files. It is important
to note that incremental hot backups can be
invalidated by use of the bulk insert optimization.
For more information, see the
section on Hot Backup in the Getting Started
With Transaction Processing Guide and the description of the flag
DB_HOTBACKUP_IN_PROGRESS
in
DB_ENV->set_flags.
The bulk insert optimization is effective only for
top-level transactions. The DB_TXN_BULK flag is ignored
when parent is non-null.
DB_TXN_NOSYNC
Do not synchronously flush the log when this transaction commits or prepares. This means the transaction will exhibit the ACI (atomicity, consistency, and isolation) properties, but not D (durability); that is, database integrity will be maintained but it is possible that this transaction may be undone during recovery.
This behavior may be set for a Berkeley DB environment using the DbEnv::set_flags() method. Any value specified to this method overrides that setting.
If a lock is unavailable for any Berkeley DB operation performed in the context of this transaction, cause the operation to return DB_LOCK_DEADLOCK (or DB_LOCK_NOTGRANTED if the database environment has been configured using the DB_TIME_NOTGRANTED flag).
This behavior may be set for a Berkeley DB environment using the DbEnv::set_flags() method. Any value specified to this method overrides that setting.
This transaction will execute with snapshot isolation. For databases with the DB_MULTIVERSION flag set, data values will be read as they are when the transaction begins, without taking read locks. Silently ignored for operations on databases with DB_MULTIVERSION not set on the underlying database (read locks are acquired).
The error DB_LOCK_DEADLOCK will be
returned from update operations if a snapshot transaction attempts to
update data which was modified after the snapshot transaction read it.
DB_TXN_SYNC
Synchronously flush the log when this transaction commits or prepares. This means the transaction will exhibit all of the ACID (atomicity, consistency, isolation, and durability) properties.
This behavior is the default for Berkeley DB environments unless the
DB_TXN_NOSYNC flag was specified to the
DbEnv::set_flags()
method.
Any value specified to this method overrides that setting.
DB_TXN_WAIT
If a lock is unavailable for any Berkeley DB operation performed in the context of this transaction, wait for the lock.
This behavior is the default for Berkeley DB environments unless the
DB_TXN_NOWAIT flag was specified to the
DbEnv::set_flags()
method.
Any value specified to this method overrides that setting.
DB_TXN_WRITE_NOSYNC
Write, but do not synchronously flush, the log when this transaction commits. This means the transaction will exhibit the ACI (atomicity, consistency, and isolation) properties, but not D (durability); that is, database integrity will be maintained, but if the system fails, it is possible some number of the most recently committed transactions may be undone during recovery. The number of transactions at risk is governed by how often the system flushes dirty buffers to disk and how often the log is flushed or checkpointed.
This behavior may be set for a Berkeley DB environment using the DbEnv::set_flags() method. Any value specified to this method overrides that setting.
The DbEnv::txn_begin()
method may fail and throw a DbException
exception, encapsulating one of the following non-zero errors, or return one
of the following non-zero errors:
The maximum number of concurrent transactions has been reached.
DbMemoryException is thrown if
your Berkeley DB API is configured to throw exceptions.
Otherwise, ENOMEM is returned.