When a server is stopping, it'll send an error to the client before
terminating the TCP connection. The code in readwritesplit would detect
this error and create a hangup event on the DCB. This would cause it to
appear as if the TCP connection was broken and the router would
immediately try to reconnect to the same server.
By ignoring the error and allowing the connection to die on its own, we
avoid immediately reconnecting and retrying any transactions on the
stopping server. This increases the chances that the monitor will see it
first and assign the server states correctly before the transaction replay
is attempted.
The assertion would hold true for a single worker but it can't be
guaranteed to be true on a multi-worker system where the statistics are
distributed across the workers.
Enabling the feature by default prevents the master connection from dying
during times when there are very little or no writes. Having a modest ping
interval of 300 seconds serves to minimize the amount of extra work that
both MaxScale and the server have to do while still keeping the
connections in good shape.
Used only with supporting server versions. Using the time limit ensures that
the server interrupts the query at the same point Connector-C would cut the
connection. This prevents lingering queries.
Also, cleans up some associated error messages.
Rewrote the bug561.sh test as the error_messages test that covers what was
tested by the script as well as some new parts that were untested. This
revealed a bug in the error messages where MaxScale always returns the
database name in the error.
When connecting to a node, a database can now be optionally given as a
parameter. This makes testing with different databases easier as the need
to use the explicit functions is removed.
RPM packages by default strip all executables on some systems after
installation. To work around this, the post install part needs to be
prevented. This does not mean the post-install scripts used to create the
directories required by MaxScale.
Fetching the users may potentially take longer than the watchdog
timeout. To ensure that MaxScale is not killed by systemd, we must
ensure that the notifications are generated also when MaxScale
synchronously is fetching the users.
maxscale-system-test/set_env.sh script is slow because it checks if MariaDB start command is mysql or mysqld
Now all versions can be started with 'mysql' except MySQL 5.5 which is not supported
Some rearrangements to ensure that what should be private
can be kept private.
- WatchdogNotifier made a friend.
- WatchdogWorkaround defined in RoutingWorker and made a friend.
- mxs::WatchdogWorker defined with 'using'.
The systemd watchdog mechanism requries notifications at
regular intervals. If a synchronous operation of some kind
is performed by a worker, then those notfications will not
be generated.
This change provides each worker with a secondary thread that
can be used for triggering those notifications when the worker
itself is busy doing other stuff. The effect is that there will
be an additional thread for each worker, but most of the time
that thread will be idle.
Sofar only the mechanism; in subsequent changes the mechanism
will be taken into use.
Any administrative operation, performed using maxadmin or maxctrl,
that takes longer than the watchdog timeout will cause MaxScale to
be killed and restarted by the watchdog.
Added a test case that does a set of sanity checks on the monitor. As the
monitor is very simple, there are not a lot of things to test without
access to the actual instances (e.g. ExeMgr failures need to be tested).
Currently the test always passes as ColumnStore clusters aren't
implemented for the test framework.
The test left undropped databases after itself which caused unnecessary
work with repeated runs. Also removed the somewhat superficial wait at the
start of the test.
Given that the pattern is given as an argument quoted with single quotes,
embedded single quotes are a problem. Given that most cases that they are
used for is as parts of natural words like can't or won't. By replacing
the single quotes with wildcard characters, we make sure the input is
formatted correctly while still matching with relatively high accuracy.
