If local address has been specified, then all connections created
using mxs_mysql_real_connect() will use that same local address as
well.
A system test has not been created as our VMs do not have more than
one usable IP-address. Locally it has been verified to work as
expected.
The test appears to fail due to connection errors when it attempts to
check whether MaxScale is still alive. To offset the chance of the backend
server still refusing connections after the initial spike, the sleep
before the check was increased.
The code had a note in that states that the test uses custom code
backported from 2.2 and that it should be updated to use common code once
merged into 2.2.
Added a small sleep to make sure the monitor picks up the changes in the
topology.
When MaxScale thinks that the master has failed, it tries to verify it by
seeing if the slave server is receiving events. There was a missing IO
thread status check in the slave_receiving_events function which caused
the failover to wait until the verification timed out.
The relay master detection logic also lacked a check for the slave SQL
thread status. The code should check the state of the SQL thread to
determine whether the server is actually a functional slave to a master.
Removed the `socket` options from the network listeners.
Updated wording for the listener sections to only mention network
ports. This makes the tutorial more concise.
The password encryption was documented in the installation instructions
and the main configuration document referred to it as a "section" in the
same document. Having it in the main configuration document makes it the
authoritative source of information.
The fix causes a regression in the failover functionality as there is a
dependency between the slave's master ID and how the failover
performs. This dependency should not exist but fixing it causes a problem
with the mysqlmon_rejoin_bad2 test.
The variable containing the number of workers must be updated
only after the workers have been successfully created.
Failure to do this led to crash in Worker::shutdown_all() if a
terminating signal was received after the worker initialization
had failed.
Instead of trying to figure out whether the kernel supports O_DIRECT
in conjunction with pipes, let's just use it and if it fails, try
without O_DIRECT.
Case in point, based on circumstantial evidence it seems that in a
container context, it may appear as if the kernel supports O_DIRECT
when it in reality does not. So better to use brute-force.
AdressSanitizer is a lightweight memory error detector that instruments at
compile time instead of at execution time. This allows serious memory
errors to be detected without the cost of slowing down the whole program
that often happens when Valgrind is used. It is also easier to enable for
test runs as it is a simple compiler flag.
This is a cherry-pick of fb9b027842a7b65c4826455cd34d88e2f5f28e79 from the
2.2 branch.
With the `allow_master_change` parameter enabled, sessions can start using
a different master node if one is available. This will not prevent
sessions from closing if a write query is received while no master
replacement is available.
If a listener section specifies both a 'socket' and a 'port' the
creation will fail with a clear error message.
If 'address' and 'socket' is specified, there will be a warning that
the address is meaningless.
It is now impossible to create two listeners for a service that
would listen on the same port/socket (as before), but the error
message is now sensible and provides detailed information to the
user.
Startup now done in a static method. Constructor initializes some values.
Config parameters loaded in a separate method. Some things still need
looking.
When MaxScale is starting, the loading of the listeners can take a while
if there are a large number of services and users to load. To signal this
to the user, progress messages should be logged after every service is
started.
