DataNucleus utilises datastore connections as follows
-
EMF : single connection at any one time for datastore-based value generation. Obtained just
for the operation, then released
-
EMF : single connection at any one time for schema-generation. Obtained just for the
operation, then released
-
EM : single connection at any one time. When in a transaction the connection is held
from the point of retrieval until the transaction commits or rolls back. The exact point at
which the connection is obtained is defined more fully below.
When used for non-transactional operations the connection is obtained just for the
specific operation.
For
pessimistic
(or datastore) transactions when
begin()
is called on the transaction,
a connection will be obtained to the datastore. This datastore connection will be held
for the
duration of the transaction
until such time as either
commit()
or
rollback()
are
called.
For
optimistic
transactions things are a little more complicated. The call to
begin()
has no effect with respect to datastore connections. All updates to the datastore are delayed until
flush() or commit() and so no connection is obtained until that time. When flush() is called, or
the transaction committed a datastore connection is finally obtained and it is held open until
commit/rollback completes. when a datastore operation is required. The connection is typically
released after performing that operation. So datastore connections, in general, are held for much
smaller periods of time.
This is complicated slightly by use of the PMF property
java.jdo.option.IgnoreCache
.
When this is set to
false
, the connection, once obtained, is not released until the call to
commit()/rollback().
If you have multiple threads using the same PersistenceManager then you can get "ConnectionInUse"
problems where another operation on another thread comes in and tries to perform something while
that first operation is still in use. This happens because the JPA spec requires an implementation
to use a single datastore connection at any one time. When this situation crops up the user ought
to use multiple EntityManagers.
Another important aspect is use of queries for Optimistic transactions, or for non-transactional
contexts. In these situations it isn't possible to keep the datastore connection open indefinitely
and so when the Query is executed the ResultSet is then read into core making the queried objects
available thereafter.
Sometimes a connection to a datastore can take some time to obtain. For that reason some
datastores can benefit from pooling of connections. Please consult the documentation for the
datastore in use for what is available, in particular for
RDBMS and
LDAP.