When a read was successfully retried, the original expected response was
not decremented from the reponse counter. This caused one extra response
to be expected for successfully retried reads which caused a hang after a
retried read.
When a session command was executed and the last slave that was executing
dies, readwritesplit would route any pending queries before closing the
slave connection. This could cause a hang if the routing logic decided to
pick the failed server as the target of the stored query.
This fixes the MXS-1323 related regression in the develop branch.
It is now possible to specify the thread stack size to be used,
when a new thread is created. This will subsequently be used
for allowing the stack size to be specified for worker threads.
Disabling the tee filter at runtime is desirable for cases where the
branched service is not always needed. Migrations and pre-production
setups are one where changes to the branched service are expected and
splitting the queries would result in an error.
All monitors now persist the state of the server in a monitor journal
file.
Moved the removal of stale journals into the core and removed them from
the monitor journal interface.
The function type rule now accepts an empty list of functions, allowing
all functions to be blocked by defining an empty function type rule and
setting action=allow.
A new option ‘slave_hostname’ allows the setting of hostname in
COM_REGISTER_SLAVE.
SHOW SLAVES HOSTS; in master server can show the hostname set in binlog
router:
MariaDB [(none)]> SHOW SLAVE HOSTS;
+-----------+-----------------------------+------+-----------+
| Server_id | Host | Port | Master_id |
+-----------+-----------------------------+------+-----------+
| 93 | maxscale-blr-1.mydomain.net | 8808 | 10124 |
+-----------+-----------------------------+------+-----------+
The qlafilter exposes the unified log file as a module command that
outputs an array of JSON strings, one for each line in the file. The
command takes two optional parameters, the start and end lines for the log
file.
Further improvements could be done to allow reading of session specific
log files.
Some sort of an authorization scheme for the REST API should be
implemented to limit the possibility of exposing unwanted parts of the
REST API.
If a destroyed monitor is created again, it will be reused. This should
prevent excessive memory growth when the same monitor is created and
destroyed again.
If a connection has not been fully established (i.e. authentication has
been completed) then it should not be considered as a connection pool
candidate.
Divided functionality into classes, fixed comments +
various other cleanup. BackenAuth no longer increments
sequence on sending password. SQLite busy timeout shortened
to 1 second.
This includes the client and backend authenticators. Currently,
only a simple password-based scheme with the SQL-client "dialog" plugin
is supported. In this mode, the server sends the first PAM message
with the AuthSwitchRequest packet and the client responds with the
password. No further authentication messages are supported. If the
connection is not encrypted, the password is sent in plaintext. The
client password is used as is for logging in to backends.
PURGE BINARY LOGS; deletes all files in binlogdir and GTID maps repo
but keeps current binlog file.
PURGE BINARY LOGS TO ‘file’; deletes all files in binlogdir and GTID
maps repo up to specified file.
mariadb10_slave_gtid=On option is needed in order to keep the list of
binlog files.
The setting parsing is now similar to the other server settings.
The header is printed if log_info is on.
Changed the setting name to simply "proxy_protocol".
Updated documentation.
If option ‘binlog_structure’ is set to ‘tree’ then SHOW BINARY LOGS
displays the tree details of the binlog files.
MySQL [(none)]> SHOW BINARY LOGS;
+--------------------------+-----------+
| Log_name | File_size |
+--------------------------+-----------+
| 0/10122/mysql-bin.000117 | 1167 |
| 0/10122/mysql-bin.000118 | 652 |
| 0/10124/foo-bin.000016 | 5082 |
| 0/10124/foo-bin.000017 | 491 |
+--------------------------+-----------+
With option set to ‘flat’ (which is the default) the output contains
only
names:
MySQL [(none)]> SHOW BINARY LOGS;
+------------------+-----------+
| Log_name | File_size |
+------------------+-----------+
| mysql-bin.000110 | 425 |
| mysql-bin.000111 | 10409 |
| mysql-bin.000112 | 9659 |
+------------------+-----------+
SHOW [FULL] BINARY LOGS is now able to report the same filename in use
with different server_ids: this can happen with binlog_structure=tree
example from SHOW FULL BINARY LOGS
0/10122/mysql-bin.000113
…
0/10122/mysql-bin.000116
…
0/5306/mysql-bin.000113
SHOW BINARY LOGS shows the same file twice:
mysql-bin.000113
…
mysql-bin.000116
…
mysql-bin.000113
The backend server can send a response even if the client hasn't sent a
request. One case where this occurs is when the server is shutting
down. The internal logic of readwritesplit can't handle unexpected states
gracefully so the safest thing to do is to just ignore them and send the
responses to the client.
Enabling the option hinders the use of maintenance mode with the root
master node in most use-cases.
This behavior occurs due to the fact that the maintenance mode causes a
server to be treted as if it was down. The Galera monitor waits for the
cluster to reorganize before assigning a new master node. This is correct
(but very unexpected) behavior for single instance use-cases.
Add Slave_pos for Using_Gtid in SHOW SLAVE STATUS only if Master GTID
mode is set
Possible output values are:
Default
Using_gtid: No
or
Using_gtid: Slave_pos
The query classifier should only be used to parse text protocol
statements. The insertstream filter exploited the fact that any statements
that the filter did not expect would be classified as an unknown
commands. This led to repetitive error messages with binary protocol
statements.
When a backend is waiting for a response but no statement is stored for
the session, the buffer where the stored statement is copied is not
modified. This means that it needs to be initialized to a NULL value.
Added a test that checks that the behavior works as expected even with
persistent connections. A second test reproduces the crash by executing
parallel SET commands while slaves are blocked.
There is still a behavioral problem in readwritesplit. If a session
command is being executed and it fails on a slave, an error is sent to the
client. In this case it would not be necessary to close the session if the
master is still alive.
The schemarouter should always use Backend::write for communication with
the backend servers. This keeps the backends in the correct state.
This fix solves some of the random test failures in the `sharding` test.
