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.
The update_names() will have to be updated to take a possible alias
name as well. That needs to be tracked inside QcSqliteInfo, so that
when there is a statement like "select t2.a from t1 t2" we report
the field as t1.a and not as t2.a.
The structure for storing statement information is now a C++ class.
At least initially it will not be turned into a full-fledged class
that would contain all functionality, but the code will be a mishmash
of C and C++.
The replication needs to be stopped before the binlogrouter is started. If
the replication is stopped after this, it is possible that two servers
with the same value of server_id attempt to register as slaves which
causes the later of them to fail.
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 replication_manager is only designed for systems that have yum
installed which means it will always fail on non-RHEL/CentOS systems.
The query threads in mxs1323_stress were not checking whether the test had
ended while they were executing the queries. This caused test timeouts as
the queries can take a relatively long time.
The tests should not drop the test database and then recreate it. This
adds an unnecessary burden on replication which will cause false positives
when replication is not fast enough.
This is commit `2410c23e6664bef7e449dfd5a767121f309559fe` from develop
cherry-picked to 2.1.
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.
As the test does a lot of large inserts, the timeouts can be triggered at
times when the test is not actually stuck but just waiting for the insert
to complete. Increasing the timeout from 10 to 30 seconds should give it
enough time to finish processing.
Earlier when the log file was flushed, it was also truncated,
which meant that the log content was log, unless the file first
was moved. Now it will only be opened. Thus, no data can be lost
due to the operation. If logrotate it used, the end result will
be the same, as the file is moved first.
