Fixes bugs found since release 5.1.12.
Bugs fixed:
Logic in implementations of
LoadBalancingConnectionProxy
and
LoadBalanceStrategy
behaved differently as to
which SQLException
s trigger failover to a new
host. The former looked at the first two characters of the
SQLState:
if (sqlState.startsWith("08")) ...
The latter used a different test:
if (sqlEx instanceof CommunicationsException || "08S01".equals(sqlEx.getSQLState())) { ...
This meant it was possible for a new
Connection
object to throw an
Exception
when the first selected host was
unavailable. This happened because
MySqlIO.createNewIO()
could throw an
SQLException
with a
SQLState
of “08001”, which did
not trigger the “try another host” logic in the
LoadBalanceStrategy
implementations, so an
Exception
was thrown after having only
attempted connecting to a single host.
(Bug#52231)
In the file DatabaseMetadata.java
, the
function private void
getCallStmtParameterTypes
failed if the parameter was
defined over more than one line by using the '\n' character.
(Bug#52167)
The catalog parameter, PARAM_CAT
, was not
correctly processed when calling for metadata with
getMetaData()
on stored procedures. This was
because PARAM_CAT
was hardcoded in the code
to NULL
. In the case where
nullcatalogmeanscurrent
was
true
, which is its default value, a crash did
not occur, but the meta data returned was for the stored
procedures from the catalog currently attached to. If, however,
nullcatalogmeanscurrent
was set to
false
then a crash resulted.
Connector/J has been changed so that when
NULL
is passed as
PARAM_CAT
it will not crash when
nullcatalogmeanscurrent
is
false
, but rather iterate all catalogs in
search of stored procedures. This means that
PARAM_CAT
is no longer hardcoded to
NULL
(see Bug#51904).
(Bug#51912)
A load balanced Connection
object with
multiple open underlying physical connections rebalanced on
commit()
, rollback()
, or
on a communication exception, without validating the existing
connection. This caused a problem when there was no pinging of
the physical connections, using queries starting with “/*
ping */”, to ensure they remained alive. This meant that
calls to Connection.commit()
could throw a
SQLException
. This did not occur when the
transaction was actually committed; it occurred when the new
connection was chosen and the driver attempted to set the
auto-commit or transaction isolation state on the newly chosen
physical connection.
(Bug#51783)
The rollback()
method could fail to rethrow a
SQLException
if the server became unavailable
during a rollback. The errant code only rethrew when
ignoreNonTxTables
was true and the exception
did not have the error code 1196,
SQLError.ER_WARNING_NOT_COMPLETE_ROLLBACK
.
(Bug#51776)
When a StatementInterceptor
was used and an
alternate ResultSet
was returned from
preProcess()
, the original statement was
still executed.
(Bug#51666)
Objects created by ConnectionImpl
, such as
prepared statements, hold a reference to the
ConnectionImpl
that created them. However,
when the load balancer picked a new connection, it did not
update the reference contained in, for example, the
PreparedStatement
. This resulted in inserts
and updates being directed to invalid connections, while commits
were directed to the new connection. This resulted in silent
data loss.
(Bug#51643)
jdbc:mysql:loadbalance://
would connect to
the same host, even though
loadBalanceStrategy
was set to a value of
random
, and multiple hosts were specified.
(Bug#51266)
User Comments
Add your own comment.