The create and destroy entry points allow authenticators to store data in
the DCB. This data is not shared by other DCBs related to the same
session.
The plugin_name entry point wasn't really useful as the plugins would
still need to send a AuthSwitchRequest packet if they wanted to change the
authentication mechanism.
The COM_QUIT packets should be sent to the backends if persistent
connections aren't used. This allows for a controlled shutdown of the
connections on both ends even if the client closes the connection before
all backends have authenticated.
The authentication for backend connections is now done in the
MySQLBackendAuth module. This is also the default authentication module
for backend connections created by MySQLBackend.
The backend responses are now read in one place and the functions just
read the data. The protocol level will now handle the packet gathering
process and the authentication part just inspects the data.
Backend connections now load authenticators when they are being
connected. In the future, this enables the use of authentication modules
for backend connection.
The DCB error messages now log the type of the DCB and the remote address
in addition to the system error message. The file descriptor and memory
address are no longer printed in the error message as they are not useful
to the end user. The fd and address are now logged at debug level with a
more verbose error message.
As the failover status check is done after the pending status has been
moved to the current status, the do_failover should set the current status
of the server as Master.
Events saved and read have 15 bytes of header XORed with IV.
Partial events are still not handled.
Next implementation will encrypt the whole event instead of 15 bytes XOR
As errors and warnings are throttled, there is a need for being able
to log severe errors without ever having them throttled (e.g. when
logging the stack in conjunction with a crash).
MXS_ALERT should only be used in a context where the process is known
to be going down, either via crash or explicit exit.
When the authentication string was decoded from hexadecimal to binary, it
was possible that an out of bounds read was done if the length of the data
was not an even number.
The storage_rocksdb version is now stored to the database. That will
ensure that should the format (key content, length, etc.) change, we
can will detect whether a database is too old and take action.
The databases targeted by a query are now included in the key.
That way, two identical queries targeting a different default
database will not clash.
Further, it will mean that queries targeting the same databases
are stored near each other, which is good for the performance.
The concept of 'allowed_references' was removed from the
documentation and the code. Now that COM_INIT_DB is tracked,
we will always know what the default database is and hence
we can create a cache key that distinguises between identical
queries targeting different default database (that is not
implemented yet in this change).
The rules for the cache is expressed using a JSON object.
There are two decisions to be made; when to store data to the
cache and when to use data from the cache. The latter is
obviously dependent on the former.
In this change, the 'store' handling is implemented; 'use'
handling will be in a subsequent change.
