The error logging is now more detailed and tells why the connection is
being closed. This should help the user figure out what is happening when
write fails and the connection is closed.
Some of the master server status checks didn't check whether the server
was actually running. The macros in server.h should always be used instead
of manually inspecting the server status.
If a readwritesplit session is active, it should never connect to a new
master. This will lead to unexpected results as the session states aren't
consistent.
If an illegal DCB close is done with a backend DCB, it will log the server
where it was connected. This allows us to know whether the DCB was
connected to a master or a slave.
Added more debug assertions to readwritesplit code. The DCBs should never
enter the DCB_STATE_DISCONNECTED.
Removed useless debug log messages. The messages usually just flood the
logs with no use to the developers.
The filter can detect SERVER_MORE_RESULTS_EXIST which means the server
is sending more result sets: example:
DROP PROCEDURE IF EXISTS multi;
DELIMITER $$
CREATE PROCEDURE multi() BEGIN
SELECT 1;
SELECT id FROM t2 limit 40;
set @a=4;
SELECT 2;
END$$
DELIMITER ;
MySQL> call multi()
A debug assertion failed due to a NULL buffer but a non-zero packet
length. This was caused by a missing reset of the packet length after
freeing the buffers.
If a Galera node has a nonpositive priority, the node will never be chosen
as the master. This gives the user more control over how the master is
chosen.
Some error messages were logged at INFO level and some had conditions that
prevent the logging. Removed these restrictions that an error situation is
always logged.
The luafilter didn't use a format string with dcb_printf which can lead to
unexpected results if the returned string contains printf special
characters.
If a master_failure_mode was set to error_on_write, a reconnection to the
old master would happen after the following events:
- Master server fails and the connection is closed
- The master server recovers
- A slave fails and the connection is closed
- A replacement for the slave is searched
If these events took place, the master would be taken back into use with
an inconsistent session state.
When a persistent connection is taken from the pool, the state is reset
with a COM_CHANGE_USER on the next write. This allows reuse of persistent
connections without having to worry about the state of the MySQL session.
Binlog server is already configured: if there is no pending transaction
a new binlog file is created after CHANGE MASTER.
If as START SLAVE is issued replication starts as usuale.
If maxscale is restarted the replication starts using the new created
file.
While configuring binlog server for the first time, master.ini not
existent, the specified MASTER_LOG_FILE is created in the $binlogdir.
If START SLAVE command is not issued the replication can start after
restarting maxscale as the binlog file exists.
When a COM_CHANGE_USER statement was executed, the new user credentials
were copied after the authentication message was sent. This caused the
COM_CHANGE_USER to always succeed the first time as it used the current
credentials. The user credentials would always lag behind by one.
When checksum is in use and there is an error in replication stream
master connection the blr_terminate_master_replication has no effect.
MXS-961: The checksum detection calls
blr_master_delayed_connect(router); and connection is scheduled again.
The fix will break the main loop as soon as the error indicator byte is
seen and no other computation will be done (such as checksum)
The MAXROWS_DISCARDING_RESPONSE is handled differently: the OK packet
is sent only after an EOF is seen in a reply even with multiple packet
transmission from the backend
max rows filter first implementation
Resultsets with more than max_resultset_rows
will be skipped and empty result set is returned to client.
Not yet tested with multi_statements
[MaxRows]
type=filter
module=maxrows
max_resultset_rows=10
debug=15
Now that a filter can express that the transaction state is tracked,
the cache implementation can be simplified. We do not need to cater
for the case that a "too short" or "too long" a packet would be
delivered.
Further, since the autocommit mode and transaction state of the session
are tracked, the filter can cache data when it is safe to do so. In
practice that means when either AUTOCOMMIT is ON and no explicit
transaction is active or when a READ ONLY transaction is active,
irrespective of the autocommit state.
In principle it would be possible to tentatively cache data during
a transaction, and if the transaction is committed successfully
flush the tentatively cached data to the actual cache, but that will
be for another day.
The transaction state only reflects explicitly started transactions.
Thus, by looking at the autocommit mode and the transaction state a
component can figure out whether the current statement will be committed
or not.
The transaction state must be updated after a buffer has been split
into buffer containing individual packets.
NOTE: The actual updating of the transaction state and the autocommit
mode is currently wrong, but will be updated in a subsequent change.
The authentication checks make sure that a user has all the required
grants to access the database. This prevents the creation of unnecessary
backend connections reducing the overall load on the database.
Doing preliminary authentication in MaxScale enables the creation of more
informative error messages.
