The order of the servers in the service definition could break the
master_accept_reads functionality.
When the first server defined in the service is a slave, it will always be
picked as the first candidate for reads. The master would only be
considered as a candidate for reads if no previous candidate was
available. For this reason, the master_accept_reads only worked when the
first server in the list was the master.
The routers no longer need to track the number of errors each DCB
receives. This is now done by the protocol modules.
The type of the DCB no longer needs to be checked in the handleError
implementation as the function is only called when a backend DCB fails.
Altered the function to assert that the DCB is a backend DCB in addition
to the existing assertions for non-NULL backend reference on function
return.
Move the fetching of the backend reference after the type of the DCB is
inspected in handleError. This removes the need to handle the case where
the returned bref is NULL and the DCB is a client DCB.
The highwater and lowwater callbacks were never registered for the client
DCBs in the binlogrouter.
The DCB hangup callbacks were never called by the core and were replaced
with fake hangup events in an earlier version.
Removed unused spinlocks from DCBs, sessions and the MySQL protocol
structs. They were used in a context where only one thread has access to
the structure.
Removed unused member variables from DCBs.
The combination of the default values of `disable_sescmd_history=false`
and `max_slave_connections=100%` does not make sense as it is not possible
to find a replacement slave in case an active one fails.
Moved some typedefs to router.h and server.h, changed a few
constants to these enums. Renamed some types in config.h to
remove "Gateway".
There are still some functions in the public header which are
only used in core, but they seem to fit the theme of public functions
so were not moved.
handleError can detect READ ONLY transaction set when problem_dcb ==
rses->forced_node->bref_dcb.
Session will be closed and rses->forced_node set to NULL
All modules now declare a name for the module. This is name is added as a
prefix to all messages logged by a module. The prefix should help
determine which part of the system logs a message.
The parameters that readwritesplit uses now use the new system. This
removes the need for the qualified parameter processing found in config.c.
All values for router_options are now also accepted as parameters. The
router_options is deprecated and support for it will be removed in a
future version.
The MXS_MODULDE object now contains optinal pointers for functions
to be called att process and thread startup and shutdown. Since the
functions were added to the end, strictly speaking, all structures
would not have needed to have been modified, but better to be
explicit. In a subsequent change, these will be called.
C++ does not support flexible arrays, so for the time being C++
modules are restricted to 10 parameters. Better approach is to
factor out the parameters to a separate array and then just store
a pointer to that array in MXS_MODULE.
The variables now use the actual option and parameter names. This should
help make the code more readable and easier to understand in relation to
the used options.
The MODULE_INFO is now the main object which is used by modules to convey
information to the MaxScale core. The MXS_MODULE name is more apt as it
now contains the actual module definition.
The old MODULES structure was moved into load_utils.c as an internal
implementation and was renamed so that it is not confused with the new
MODULE structure.
The modules are now declared with a common macro. This allows future
additions to the module loading process while also making the loaded
symbol name a constant.
This allows modules to only expose one entry point with a consistent
signature. In the future, this could be used to implement declarations of
module parameters.
gwbuf_clone cloned only the first buffer of a chain of buffers,
which never can be the desired outcome, while gwbuf_clone_all
cloned all buffers.
Now, gwbuf_clone behaves the way gwbuf_clone_all used to behave
and gwbuf_clone_all has been removed.
Readwritesplit should route all queries from cloned sessions to the
master. This allows batch statements to be safely routed.
Native readwritesplit sessions only support batched writes as batched
reads aren't very common. Once readwritesplit supports batched reads, the
special handling for cloned DCBs can be removed.
If a slave fails while a non-critical read is being executed, the read is
retried on a different server. This is controlled by the new
`retry_failed_reads` option.
Only selects done that are done outside of a transaction and with
autocommit enabled are retried.
When a DCB error occurs, the handleError entry point of the routers is
called. The caller of this entry point expects that the error handler
marks the DCB as handled. The aforementioned behavior is wrong as the
error handler should not keep track of whether the handler was already
called.
Doing batch inserts though readwritesplit would stall due to the fact that
pending session commands were stored instead of executed immediately.
Session command responses that weren't complete also discarded the partial
event instead of storing it for later use.
Closing the DCB and the backend reference that uses it at the same time
makes the error handling code clearer and removes some of the assumptions
that the code made. It will cause the DCB to be closed in multiple places
but the logic of why a DCB is being closed is more visible from the code.
This change should remove all cases where a DCB is closed without a
tightly coupled backend reference.
