This is the first in a series of commits that introduce components
that allow filters to be tested without MaxScale or backends.
Obviously the environment is not suited for all cases, but allow e.g.
the behaviour of the firewall filter to be tested in isolation.
The environment consists of helper classes loading and manipulating
modules and filters and mock-classes that provide the needed
scaffolding around a filter.
Currently all components exist under the test directory of the firewall
filter, but once proven they can be moved to a more general purpose
location.
The setting limits the maximum time a MASTER_GTID_WAIT-function
can wait. To work around this limitation, the function is now called
in a loop such that the total timeout is approximately equal to
the requested timeout.
Slave redirection is a special case, as there the total failure
is only known after all redirects have been attempted. In the
failure case, all errors from connections are gathered to one
message.
By using an the value of a global incremented integer as the
initial version number of a Dbfw instance it is ensured that every
such instance does not "inherit" any rules from a previous instance
that happened to exist in the same place as the new instance to be
created.
The error message was not 100% accurate about the value. In addition to
that, neither the value itself nor the monitor or parameter names were
printed in the error message.
The state of the monitored servers is only persisted if the states of the
servers have changed. This removes the unnecessary disk IO caused by the
writing on the monitor journal.
Adds an option 'reply_time' to log_data. When enabled, query execution time
in milliseconds is logged. Execution time is measured as the time between
routeQuery and the first clientReply.
Sending an error to the client allows the connector to show more
information to the user when the DCB is closed due to a reason internal to
MaxScale.
The error message states that the connection was killed by MaxScale to
distinct it from the error sent by the server. The error number and SQL
state are still the same as both errors should be treated the same way.
The test expected the wrong error message to be in the log. To better cope
with changes in the text, the test now uses a regular expression to do the
matching.
If a server goes down and it has the stale master bit enabled, all other
bits for the server are cleared. This allows failed masters that have been
replaced to be first detected and then reintroduced into the replication
topology.
The new `ssl_verify_peer_certificate` parameter controls whether the peer
certificate is verified. This allows self-signed certificates to be
properly used with MaxScale.
Only if the filter is in blocking mode, will a match cause an error
to be returned. Thus, no point in creating the error message unless
the filter is in blocking mode.
Whether or not a match causes the query to be blocked or to be allowed
depends on whether the filter is in white-listing or black-listing mode,
so better if the text is neutral.
Using 'not_function' it is possible to match functions other than
a specific set of ones. That will make it significantly easier to
allow certain functions to be used with certain columns.
The special handling of no arguments to `function` in conjunction
with an allowing filter has been removed. The same effect can now
be achieved, without special handling, using `not_function`, no
arguments and a blocking filter.
Implementation will follow in a subsequent commit.
The test is composed of a few parts.
1: Test that failover happens on master failure.
2: Test that a server with slave sql thread stopped is not promoted.
3: Test that a server with log_slave_updates=1 is promoted before others.
