Cleaned up the headers, removed unused structures. Changed some members to
strings instead of char arrays. Switch to router templates should now be
easier.
The sharding implementation now uses a class to abstract the details of
the shard. This allows for different design where each session makes a
copy of the global shard map which is then used for the duration of the
session. In addition to making the desing a bit clearer to understand, it
also removes lock competition between threads.
Due to the change to C++, the main entry points need to be wrapped in the
exception-safety macros. The next step in the refactoring will be to use
the router template. This will remove the need to manually define them.
The schemarouter now uses the new session commands. It uses a standard
library container to manage the execution and storage of session commands.
The session command history is disabled until a more complete refactoring
can be done.
Some of the protocol modules use ssize_t instead of size_t.
Split the function that counts the number of response packets a session
command will receive into two parts. This allows it to be reused
elsewhere.
The binlogrouter error handling closed the DCB twice. This was caused by
the change in the way the DCB error handling is done.
The protocol modules now also call the error handling routine even if the
router session is NULL. This enables the binlogrouter to manage
authentication failures correctly instead of trying to reconnect again.
General code cleanup. Routing error detection now more robust.
Remove some unused code. Debug messages now use "unique_name" when
referring to servers.
Supports hint types:
-master
-slave
-named server
-all
A default action, which is performed when no hint exists or on error,
can be set. The different actions are analogous to the hint types.
A maximum connection number for slaves can be set. If more slaves are
configured for the service, the filter will rotate slaves for new sessions.
Within a session with multiple slaves, the "route_to_slave"-hint will
also rotate among the slave backends.
The readwritesplit now sends COM_PING queries to backend servers that have
been idle for too long. The option is configured with the
`connection_keepalive` parameter.
It is now possible to specify what information the caller is interested
in. With this the cost for collecting information during the query parsing
that nobody is interested in can be avoided.
The match data needs to be unique for each thread, so for the time
being it is created whenever it is needed. A more performant (although
possibly to a negigible amount) solution would be to have a separate
match data for each thread, but that will have to wait for 2.2.
When MaxScale is being started and the users are loaded, the MySQL
authenticator should not load the database users for internal services
abstracted as servers.
The loading of users at startup for internal services is avoided because
the startup is done in a single thread context and the internal services
have not yet been started.
The delayed loading of users will cause the authentication to fail when
the first client connect. This triggers the reloading of the users and the
second attempt at authentication will succeed. All of this is hidden from
the end user.
If a server points to a local MaxScale listener, the permission checks for
that server are skipped. This allows permission checks to be used with a
mix of external servers and internal services.
The static module capabilities are now used to query the capabilities of
filters and routers. The new RCAP_TYPE_NOAUTH capability is also taken
into use. These changes removes the need for the `is_internal_service`
function.
The static capabilities declared in getCapabilities allows certain
capabilities to be queried before instances are created. The intended use
of this capability is to remove the need for the `is_internal_service`
function.
The filter now checks what types of parameters it has been given and
only accepts the old style (match, server) XOR new style (match01,
target01 etc).
Adds support for special targets ->master and ->slave.
Code cleanup.
- Selects are picked out using custom parsing, so if a statement is
anything else but a SELECT, the cache will never cause the statement
to be parsed.
- The setting of of the cache parameter `selects` is taken into account.
If it is `assume_cacheable` then the statement will also not be parsed
even if it is a SELECT.
The server state information should be persisted even if a controlled
shutdown is done. This will allow the monitor to retain the server state
information across a restart.
The MySQL monitor stores the server states in a backup file which can be
used to restore the state of the servers even if MaxScale is stoppen in an
uncontrolled fashion.
The hintrouter is now in principle capable of routing requests
to the master or to some slave (in a round robin fashion) based
upon hints set by some earlier filter.
Note that as the router is completely oblivious of transaction
boundaries, using it with transactions and autocommit being off
will not make anyone happy.
Recognizing transaction boundaries using regexes and then pinning
the server until transaction commit would be needed.
The original approach was made for RocksDB where it is beneficial
to keep keys of stuff related to each other close to each other.
However, as RocksDB is no longer the primary focus, it just causes
additional cost to dig out the table names.
The key is a 64-bit integer, but crc32 only gives us a 32-bit one.
We create an other 32-bit value by running crc32 over the same SQL,
using the first crc value as adler.
I think that further reduces the chance for clashes:
uint32_t crc0 = crc32(0, Z_NULL, 0);
uint32_t crc1;
uint32_t crc2;
crc1 = crc32(crc0, "codding", 7) => 1774765869
crc2 = crc32(crc1, "codding", 7) => 1409592046
crc1 = crc32(crc0, "gnu", 3) => 1774765869
crc2 = crc32(crc1, "gnu", 3) => 1213798908
Note that the first value is the same, but the second is not.
